1、集成车牌识别模型，支持车牌检测与识别

2、新增 Milvus 身份验证支持
3、目标检测功能升级：可指定类别及topk
4、支持自定义线程池线程数量

README.md

@@ -192,6 +192,25 @@ SmartJavaAI是专为JAVA 开发者打造的一个功能丰富、开箱即用的
         </div>
       </td>
     </tr>
+    <tr>
+      <td>
+        <div align="left">
+          <p>车牌识别</p>
+          - 单层/双层检测 <br>
+          - 车牌颜色识别 <br>
+        </div>
+      </td>     
+      <td>
+        <div align="center">
+        <img src="https://cdn.jsdelivr.net/gh/geekwenjie/SmartJavaAI-Site/images/ocr/plate_recognized.jpg" width = "500px"/>
+        </div>
+      </td>
+<td>
+        <div align="center">
+        <img src="https://cdn.jsdelivr.net/gh/geekwenjie/SmartJavaAI-Site/images/ocr/plate_recognized2.jpg" width = "500px"/>
+        </div>
+      </td>
+    </tr>
     <tr>
       <td>
         <div align="left">
@@ -251,21 +270,22 @@ SmartJavaAI是专为JAVA 开发者打造的一个功能丰富、开箱即用的
   - 支持任意角度识别，方向校准
   - 支持通用文字识别，通用手写字识别
   - 支持表格识别
+  - 支持中文车牌识别：单层/双层检测，颜色识别，支持12种中文车牌
 - **机器翻译**
   - 集成NLLB-200模型：支持200+语言互相翻译
 
 
 ## 🌟 AI集成方式对比
 
-| 方案                | 技术特点                                                                 | 优点                                                                | 缺点                                                                |
-|---------------------|--------------------------------------------------------------------------|---------------------------------------------------------------------|---------------------------------------------------------------------|
-| **OpenCV**         | 传统图像处理方案                                                        | ✅ 提供java接口<br>✅ 轻量级部署<br>✅ 社区资源丰富                  | ❌ 基于传统算法精度低(60%-75%)<br>❌ 需本地安装环境 |
-| &zwnj;**商业闭源SDK（如虹软等）**&zwnj;       | 商业级闭源解决方案                                                      | ✅ 开箱即用<br>✅ 提供完整文档和SDK<br>✅ 支持离线活体检测      | ❌ 免费版需年度授权更新<br>❌ 商业授权费用高<br>❌ 代码不可控 |
-| &zwnj;**云API(阿里云)**&zwnj;   | SaaS化云端服务                                                          | ✅ 零部署成本<br>✅ 支持高并发<br>✅ 自带模型迭代        | ❌ 网络延迟风险(200-800ms)<br>❌ 按调用量计费<br>❌ 有数据安全风险 |
+| 方案                | 技术特点                                                                 | 优点                                                                | 缺点                                            |
+|---------------------|--------------------------------------------------------------------------|---------------------------------------------------------------------|-----------------------------------------------|
+| **OpenCV**         | 传统图像处理方案                                                        | ✅ 提供java接口<br>✅ 轻量级部署<br>✅ 社区资源丰富                  | ❌ 基于传统算法精度低(60%-75%)<br>❌ 需本地安装环境             |
+| &zwnj;**商业闭源SDK（如虹软等）**&zwnj;       | 商业级闭源解决方案                                                      | ✅ 开箱即用<br>✅ 提供完整文档和SDK<br>✅ 支持离线活体检测      | ❌ 免费版需年度授权更新<br>❌ 商业授权费用高<br>❌ 代码不可控          |
+| &zwnj;**云API(阿里云)**&zwnj;   | SaaS化云端服务                                                          | ✅ 零部署成本<br>✅ 支持高并发<br>✅ 自带模型迭代        | ❌ 网络延迟风险(200-800ms)<br>❌ 按调用量计费<br>❌ 有数据安全风险  |
 | &zwnj;**Python混合调用**&zwnj;  | 跨语言调用方案                                                          | ✅ 可集成PyTorch/TF等框架<br>✅ 支持自定义算法<br>✅ 识别精度高  | ❌ 需维护双语言环境<br>❌ 进程通信性能损耗(30%+)<br>❌ 异常处理复杂度翻倍 |
-| &zwnj;**JNI/JNA**&zwnj;         | 跨语言底层调用方案      | ✅ 直接调用 C/C++ 高性能算法库✅ 支持调用各种原生成熟库✅ 可封装成通用工具Jar   |❌ 开发成本高，JNI更复杂❌ 跨平台兼容性差       |
-| &zwnj;**DJL框架**&zwnj;         | 深度学习框架                                                            | ✅ 纯Java实现<br>✅ 支持主流深度学习框架<br>✅ 可加载预训练模型(99%+)   | ❌ 需掌握DL知识<br>❌ 需处理模型加载、预处理、后处理等复杂技术细节      |
-| &zwnj;**SmartJavaAI**&zwnj;       | java深度学习工具包                                                            | ✅ 支持主流深度学习框架<br>✅ 提供丰富、开箱即用API<br>  ✅ 上手简单，单一Jar包集成 |   ❌要求JDK版本11及以上  |
+| &zwnj;**JNI/JNA**&zwnj;         | 跨语言底层调用方案      | ✅ 直接调用 C/C++ 高性能算法库✅ 支持调用各种原生成熟库✅ 可封装成通用工具Jar   | ❌ 开发成本高，JNI更复杂❌ 跨平台兼容性差                       |
+| &zwnj;**DJL框架**&zwnj;         | 深度学习框架                                                            | ✅ 纯Java实现<br>✅ 支持主流深度学习框架<br>✅ 可加载预训练模型(99%+)   | ❌ 需掌握DL知识<br>❌ 需处理模型加载、预处理、后处理等复杂技术细节         |
+| &zwnj;**SmartJavaAI**&zwnj;       | java深度学习工具包                                                            | ✅ 支持主流深度学习框架<br>✅ 提供丰富、开箱即用API<br>  ✅ 上手简单，单一Jar包集成 | 无                                             |
 
 
 
@@ -300,7 +320,7 @@ SmartJavaAI是专为JAVA 开发者打造的一个功能丰富、开箱即用的
 <dependency>
     <groupId>cn.smartjavaai</groupId>
     <artifactId>smartjavaai-all</artifactId>
-    <version>1.0.20</version>
+    <version>1.0.22</version>
 </dependency>
 ```
 ### 3、完整示例代码
@@ -406,16 +426,6 @@ SmartJavaAI是专为JAVA 开发者打造的一个功能丰富、开箱即用的
 - 自训练模型推理
 - yolov3~yolov12 系列
 
-**SSD 系列**
-
-| 模型名称 | 引擎               | 骨干网络 | 输入尺寸 | <div style="width: 60pt">训练数据集</div> | 精度（mAP） | <div style="width: 50pt">推理速度</div> | <div style="width: 150pt">适用场景</div>|
-| :-------- |------------------| ------------- | ----------- |--------------------------------------| -------------|-------------------------------------| -------------|
-|SSD_300_RESNET5| PyTorch          | ResNet‑50 | 300×300 | COCO                                 | 中等 | 快                                   | 精度需求一般|
-|SSD_512_RESNET50_V1_VOC| PyTorch | ResNet‑50 | 512×512 | Pascal VOC                           | 稍高 | 中等                                  | 精度优先、可接受略低速度的场景|
-|SSD_512_VGG16_ATROUS_COCO| MXNet | VGG‑16 | 512×512 | COCO                                 | 较高 | 中等                                  | 通用场景；对小目标有一定提升|
-|SSD_300_VGG16_ATROUS_VOC| MXNet | VGG‑16 | 300×300 | Pascal VOC                           | 中等偏上 | 快                                   | VOC 数据集同类任务；资源受限时使用|
-|SSD_512_MOBILENET1_VOC| MXNet | MobileNet‑1.0 | 512×512 | Pascal VOC                           | 中等 | 快                                   | 嵌入式/移动端设备；算力和内存都很有限|
-
 **YOLO 系列**
 
 |模型名称 | 引擎          | 版本 | 大小（Backbone） | <div style="width: 60pt">数据集</div> | <div style="width: 50pt">精度</div> | <div style="width: 50pt">速度</div> | <div style="width: 150pt">适用场景</div> |
@@ -427,6 +437,18 @@ SmartJavaAI是专为JAVA 开发者打造的一个功能丰富、开箱即用的
 |YOLOV5S_ONNXRUNTIME| OnnxRuntime     | v5  | 小型 | COCO                                   | 较高                                | 加速（需 ONNX 支持） | Windows/Linux 通用加速部署                 |
 |YOLO (MXNet / 通用模型) | MXNet     | v3  | DarkNet‑53 | COCO                                   | 较高                                | 快 | 需要 MXNet 生态或复现老项目时使用                 |
 
+
+**SSD 系列**
+
+| 模型名称 | 引擎               | 骨干网络 | 输入尺寸 | <div style="width: 60pt">训练数据集</div> | 精度（mAP） | <div style="width: 50pt">推理速度</div> | <div style="width: 150pt">适用场景</div>|
+| :-------- |------------------| ------------- | ----------- |--------------------------------------| -------------|-------------------------------------| -------------|
+|SSD_300_RESNET5| PyTorch          | ResNet‑50 | 300×300 | COCO                                 | 中等 | 快                                   | 精度需求一般|
+|SSD_512_RESNET50_V1_VOC| PyTorch | ResNet‑50 | 512×512 | Pascal VOC                           | 稍高 | 中等                                  | 精度优先、可接受略低速度的场景|
+|SSD_512_VGG16_ATROUS_COCO| MXNet | VGG‑16 | 512×512 | COCO                                 | 较高 | 中等                                  | 通用场景；对小目标有一定提升|
+|SSD_300_VGG16_ATROUS_VOC| MXNet | VGG‑16 | 300×300 | Pascal VOC                           | 中等偏上 | 快                                   | VOC 数据集同类任务；资源受限时使用|
+|SSD_512_MOBILENET1_VOC| MXNet | MobileNet‑1.0 | 512×512 | Pascal VOC                           | 中等 | 快                                   | 嵌入式/移动端设备；算力和内存都很有限|
+
+
 **YOLOv3 变体系列**
 
 |模型名称| 引擎  | 骨干网络 |  <div style="width: 60pt">数据集</div> | 输入尺寸 | <div style="width: 50pt">精度</div> | <div style="width: 50pt">速度</div> | <div style="width: 200pt">适用场景</div> |
@@ -486,6 +508,21 @@ SmartJavaAI是专为JAVA 开发者打造的一个功能丰富、开箱即用的
 | SLANet_plus   | （增强版）该模型通过轻量级骨干 PP-LCNet、CSP-PAN 融合与 SLA Head 解码，有效提升表格结构识别的精度与速度。 | [Github](https://paddlepaddle.github.io/PaddleOCR/v3.1.0/version3.x/module_usage/table_structure_recognition.html#_3) |
 
 
+**车牌检测模型(License Plate Detection)**
+
+| 模型名称   | 模型简介                 | 模型开源网站     |
+|--------|----------------------|------------|
+| YOLOV5 | 基于YOLOV5训练，支持12种中文车牌 | [Github](https://github.com/we0091234/Chinese_license_plate_detection_recognition) |
+| YOLOV7 | 基于YOLOV7训练，支持12种中文车牌 | [Github](https://github.com/we0091234/Chinese_license_plate_detection_recognition) |
+
+
+**车牌识别模型(License Plate Recognition)**
+
+| 模型名称   | 模型简介       | 模型开源网站     |
+|--------|------------|------------|
+| PLATE_REC_CRNN | CRNN中文字符识别 | [Github](https://github.com/Sierkinhane/CRNN_Chinese_Characters_Rec) |
+
+
 ---
 
 #### 机器翻译模型
@@ -519,7 +556,22 @@ SmartJavaAI是专为JAVA 开发者打造的一个功能丰富、开箱即用的
 
 ## 近期更新日志
 
-## [v1.0.20] - 2025-07-06
+## [v1.0.22] - 2025-07-28
+- 新增 Milvus 身份验证支持
+- 集成车牌识别模型，支持车牌检测与识别
+- 目标检测功能升级：可指定类别及topk
+- 支持自定义线程池线程数量
+
+
+## [v1.0.20] - 2025-07-18
+- OCR：新增表格识别模型
+- OCR：新增9个通用模型
+- OCR：支持批量检测识别
+- OCR：新增更多参数，使用更加灵活
+- 人脸识别：支持ID查询及分页获取人脸信息
+- 活体检测：视频检测支持设置最大帧数
+
+## [v1.0.19] - 2025-07-06
 - 人脸模块：新增小视科技（MiniVision）活体检测模型
 - 人脸模块：新增阿里通义工作室活体检测模型
 - 人脸模块：新增 2 个表情识别模型

examples/face-example/pom.xml

@@ -12,7 +12,7 @@
         <maven.compiler.source>11</maven.compiler.source>
         <maven.compiler.target>11</maven.compiler.target>
         <project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
-        <smartjavaai.version>1.0.20</smartjavaai.version>
+        <smartjavaai.version>1.0.22</smartjavaai.version>
         <!--如果打包运行，需要替换成你的main-->
         <exec.mainClass>smartai.examples.face.facedet.FaceDetDemo</exec.mainClass>
 
@@ -95,6 +95,13 @@
         <dependency>
             <groupId>cn.smartjavaai</groupId>
             <artifactId>smartjavaai-face</artifactId>
+            <version>1.0.22</version>
+        </dependency>
+
+        <dependency>
+            <groupId>cn.smartjavaai</groupId>
+            <artifactId>smartjavaai-all</artifactId>
+            <version>1.0.22</version>
         </dependency>
 
 

examples/face-example/src/main/java/smartai/examples/face/VideoDemo.java

@@ -0,0 +1,25 @@
+package smartai.examples.face;
+
+import cn.smartjavaai.common.utils.VideoUtils;
+import org.bytedeco.ffmpeg.global.avcodec;
+import org.bytedeco.javacv.FFmpegFrameGrabber;
+import org.bytedeco.javacv.FFmpegFrameRecorder;
+
+/**
+ * 视频预处理
+ * @author dwj
+ * @date 2025/7/17
+ */
+public class VideoDemo {
+
+    public static void main(String[] args) {
+        try {
+            VideoUtils.rotateVideo("/Users/wenjie/Downloads/girl.mp4", "/Users/wenjie/Downloads/girl_rotate.mp4", 180,"mp4", avcodec.AV_CODEC_ID_H264);
+        } catch (FFmpegFrameRecorder.Exception e) {
+            throw new RuntimeException(e);
+        } catch (FFmpegFrameGrabber.Exception e) {
+            throw new RuntimeException(e);
+        }
+    }
+
+}

examples/face-example/src/main/java/smartai/examples/face/attribute/FaceAttributeDetDemo.java

@@ -57,7 +57,8 @@ public class FaceAttributeDetDemo {
      */
     @Test
     public void testFaceAttributeDetect(){
-        try (FaceAttributeModel faceAttributeModel = getFaceAttributeModel()){
+        try {
+            FaceAttributeModel faceAttributeModel = getFaceAttributeModel();
             DetectionResponse detectionResponse = faceAttributeModel.detect("src/main/resources/iu_1.jpg");
             //绘制并导出人脸属性图片，小人脸仅有人脸框
             BufferedImage image = ImageIO.read(new File(Paths.get("src/main/resources/iu_1.jpg").toAbsolutePath().toString()));
@@ -73,7 +74,8 @@ public class FaceAttributeDetDemo {
      */
     @Test
     public void testFaceAttributeDetect2(){
-        try (FaceAttributeModel faceAttributeModel = getFaceAttributeModel()){
+        try {
+            FaceAttributeModel faceAttributeModel = getFaceAttributeModel();
             FaceAttribute faceAttribute = faceAttributeModel.detectTopFace("src/main/resources/iu_1.jpg");
             log.info("人脸属性检测结果：{}", JSONObject.toJSONString(faceAttribute));
         } catch (Exception e) {
@@ -86,7 +88,8 @@ public class FaceAttributeDetDemo {
      */
     @Test
     public void testFaceAttributeDetect3(){
-        try (FaceAttributeModel faceAttributeModel = getFaceAttributeModel()){
+        try {
+            FaceAttributeModel faceAttributeModel = getFaceAttributeModel();
             FaceAttribute faceAttribute = faceAttributeModel.detectTopFace("src/main/resources/iu_1.jpg");
             log.info("人脸属性检测结果：{}", JSONObject.toJSONString(faceAttribute));
         } catch (Exception e) {
@@ -103,8 +106,9 @@ public class FaceAttributeDetDemo {
      */
     @Test
     public void testFaceAttributeDetect4(){
-        try (FaceDetModel faceDetModel = getFaceDetModel();
-                FaceAttributeModel faceAttributeModel = getFaceAttributeModel()){
+        try {
+            FaceDetModel faceDetModel = getFaceDetModel();
+            FaceAttributeModel faceAttributeModel = getFaceAttributeModel();
             //人脸检测
             BufferedImage image = ImageIO.read(new File(Paths.get("src/main/resources/iu_1.jpg").toAbsolutePath().toString()));
             R<DetectionResponse> detectionResponse = faceDetModel.detect(image);

examples/face-example/src/main/java/smartai/examples/face/expression/ExpressionRecDemo.java

@@ -88,7 +88,8 @@ public class ExpressionRecDemo {
      */
     @Test
     public void testExpressionDetect() {
-        try (ExpressionModel model = getExpressionModel()){
+        try {
+            ExpressionModel model = getExpressionModel();
             R<ExpressionResult> result = model.detectTopFace("src/main/resources/emotion/happy.png");
             if(result.isSuccess()){
                 log.info("识别结果：{}", JSONObject.toJSONString(result.getData().getExpression().getDescription()));
@@ -106,7 +107,8 @@ public class ExpressionRecDemo {
      */
     @Test
     public void testExpressionDetect2() {
-        try (ExpressionModel model = getExpressionModel()){
+        try {
+            ExpressionModel model = getExpressionModel();
             R<DetectionResponse> result = model.detect("src/main/resources/emotion/happy.png");
             if(result.isSuccess()){
                 //log.info("识别结果：{}", JSONObject.toJSONString(result.getData()));
@@ -128,8 +130,9 @@ public class ExpressionRecDemo {
      */
     @Test
     public void testExpressionDetect3() {
-        try (FaceDetModel faceDetModel = getFaceDetModel();
-                ExpressionModel model = getExpressionModel()){
+        try {
+            FaceDetModel faceDetModel = getFaceDetModel();
+            ExpressionModel model = getExpressionModel();
             BufferedImage image = ImageIO.read(new File(Paths.get("src/main/resources/emotion/happy.png").toAbsolutePath().toString()));
             R<DetectionResponse> detResult = faceDetModel.detect(image);
             if(detResult.isSuccess()){
@@ -156,8 +159,9 @@ public class ExpressionRecDemo {
      */
     @Test
     public void testExpressionDetect4() {
-        try (FaceDetModel faceDetModel = getFaceDetModel();
-             ExpressionModel model = getExpressionModel()){
+        try {
+            FaceDetModel faceDetModel = getFaceDetModel();
+            ExpressionModel model = getExpressionModel();
             BufferedImage image = ImageIO.read(new File(Paths.get("src/main/resources/emotion/happy.png").toAbsolutePath().toString()));
             R<DetectionResponse> detResult = faceDetModel.detect(image);
             if(detResult.isSuccess()){
@@ -182,7 +186,8 @@ public class ExpressionRecDemo {
      */
     @Test
     public void testExpressionDetectAndDraw(){
-        try (ExpressionModel model = getExpressionModel()){
+        try {
+            ExpressionModel model = getExpressionModel();
             BufferedImage image = ImageIO.read(new File(Paths.get("src/main/resources/emotion/surprise.png").toAbsolutePath().toString()));
             R<DetectionResponse> result = model.detect(image);
             if(result.isSuccess()){
@@ -206,7 +211,8 @@ public class ExpressionRecDemo {
      */
     @Test
     public void testExpressionDetectCamera(){
-        try (ExpressionModel expressionModel = getExpressionModel()){
+        try {
+            ExpressionModel expressionModel = getExpressionModel();
             OpenCV.loadShared();
             VideoCapture capture = new VideoCapture(0);
             if (!capture.isOpened()) {

examples/face-example/src/main/java/smartai/examples/face/facedet/FaceDetDemo.java

@@ -82,7 +82,8 @@ public class FaceDetDemo {
      */
     @Test
     public void testFaceDetect(){
-        try (FaceDetModel faceModel = FaceDetModelFactory.getInstance().getModel()) {
+        try {
+            FaceDetModel faceModel = FaceDetModelFactory.getInstance().getModel();
             R<DetectionResponse> detectedResult = faceModel.detect(imgPath);
             if(detectedResult.isSuccess()){
                 log.info("人脸检测结果：{}", JSONObject.toJSONString(detectedResult.getData()));
@@ -100,7 +101,8 @@ public class FaceDetDemo {
      */
     @Test
     public void testFaceDetectCustomConfig(){
-        try (FaceDetModel faceModel = getFaceDetModel()){
+        try {
+            FaceDetModel faceModel = getFaceDetModel();
             R<DetectionResponse> detectedResult = faceModel.detect(imgPath);
             if(detectedResult.isSuccess()){
                 log.info("人脸检测结果：{}", JSONObject.toJSONString(detectedResult.getData()));
@@ -118,7 +120,8 @@ public class FaceDetDemo {
      */
     @Test
     public void testFaceDetectAndDraw(){
-        try (FaceDetModel faceModel = getFaceDetModel()){
+        try {
+            FaceDetModel faceModel = getFaceDetModel();
             faceModel.detectAndDraw("src/main/resources/largest_selfie.jpg","output/largest_selfie_detected.png");
         } catch (Exception e) {
             throw new RuntimeException(e);
@@ -131,7 +134,8 @@ public class FaceDetDemo {
      */
     @Test
     public void testFaceDetectAndDraw2(){
-        try (FaceDetModel faceModel = getFaceDetModel()){
+        try {
+            FaceDetModel faceModel = getFaceDetModel();
             BufferedImage image = null;
             String imagePath = "src/main/resources/largest_selfie.jpg";
             image = ImageIO.read(new File(Paths.get(imagePath).toAbsolutePath().toString()));
@@ -154,11 +158,12 @@ public class FaceDetDemo {
      */
     @Test
     public void testDetectFaceOffine(){
-        FaceDetConfig config = new FaceDetConfig();
-        config.setModelEnum(FaceDetModelEnum.RETINA_FACE);//人脸模型
-        //模型路径,不同模型下载路径请参看文档
-        config.setModelPath("/Users/xxx/Documents/develop/face_model/retinaface.pt");
-        try (FaceDetModel faceModel = FaceDetModelFactory.getInstance().getModel(config)) {
+        try {
+            FaceDetConfig config = new FaceDetConfig();
+            config.setModelEnum(FaceDetModelEnum.RETINA_FACE);//人脸模型
+            //模型路径,不同模型下载路径请参看文档
+            config.setModelPath("/Users/xxx/Documents/develop/face_model/retinaface.pt");
+            FaceDetModel faceModel = FaceDetModelFactory.getInstance().getModel(config);
             R<DetectionResponse> detectedResult = faceModel.detect(imgPath);
             if(detectedResult.isSuccess()){
                 log.info("人脸检测结果：{}", JSONObject.toJSONString(detectedResult.getData()));
@@ -175,10 +180,11 @@ public class FaceDetDemo {
      */
     @Test
     public void testDetectFaceGPU(){
-        FaceDetConfig config = new FaceDetConfig();
-        config.setModelEnum(FaceDetModelEnum.RETINA_FACE);//人脸模型
-        config.setDevice(DeviceEnum.GPU);
-        try (FaceDetModel faceModel = FaceDetModelFactory.getInstance().getModel(config)) {
+        try {
+            FaceDetConfig config = new FaceDetConfig();
+            config.setModelEnum(FaceDetModelEnum.RETINA_FACE);//人脸模型
+            config.setDevice(DeviceEnum.GPU);
+            FaceDetModel faceModel = FaceDetModelFactory.getInstance().getModel(config);
             R<DetectionResponse> detectedResult = faceModel.detect(imgPath);
             if(detectedResult.isSuccess()){
                 log.info("人脸检测结果：{}", JSONObject.toJSONString(detectedResult.getData()));
@@ -196,7 +202,8 @@ public class FaceDetDemo {
      */
     @Test
     public void testFaceDetectSeetaface6(){
-        try (FaceDetModel faceModel = getSeetaface6DetModel()){
+        try {
+            FaceDetModel faceModel = getSeetaface6DetModel();
             R<DetectionResponse> detectedResult = faceModel.detect(imgPath);
             if(detectedResult.isSuccess()){
                 log.info("人脸检测结果：{}", JSONObject.toJSONString(detectedResult.getData()));
@@ -215,7 +222,8 @@ public class FaceDetDemo {
      */
     @Test
     public void testDetectCamera(){
-        try (FaceDetModel faceModel = getFaceDetModel()){
+        try {
+            FaceDetModel faceModel = getFaceDetModel();
             OpenCV.loadShared();
             VideoCapture capture = new VideoCapture(0);
             if (!capture.isOpened()) {

examples/face-example/src/main/java/smartai/examples/face/facerec/FaceRecDemo.java

@@ -95,6 +95,8 @@ public class FaceRecDemo {
         MilvusConfig vectorDBConfig = new MilvusConfig();
         vectorDBConfig.setHost("127.0.0.1");
         vectorDBConfig.setPort(19530);
+        //vectorDBConfig.setUsername("root");
+        //vectorDBConfig.setPassword("Milvus");
         //vectorDBConfig.setCollectionName("face5");
         //ID策略：自动生成
         vectorDBConfig.setIdStrategy(IdStrategy.AUTO);
@@ -135,7 +137,8 @@ public class FaceRecDemo {
      */
     @Test
     public void testExtractFeatures(){
-        try (FaceRecModel faceRecModel = getFaceRecModel()){
+        try {
+            FaceRecModel faceRecModel = getFaceRecModel();
             //提取图片中所有人脸特征
             R<DetectionResponse> faceResult  = faceRecModel.extractFeatures("src/main/resources/iu_1.jpg");
             if(faceResult.isSuccess()){
@@ -158,7 +161,8 @@ public class FaceRecDemo {
      */
     @Test
     public void featureComparison(){
-        try (FaceRecModel faceRecModel = getFaceRecModel()){
+        try {
+            FaceRecModel faceRecModel = getFaceRecModel();
             //基于图像直接比对人脸特征
             R<Float> similarResult = faceRecModel.featureComparison("src/main/resources/iu_1.jpg","src/main/resources/iu_2.jpg");
             if(similarResult.isSuccess()){
@@ -183,7 +187,8 @@ public class FaceRecDemo {
      */
     @Test
     public void featureComparison2(){
-        try (FaceRecModel faceRecModel = getFaceRecModel()){
+        try {
+            FaceRecModel faceRecModel = getFaceRecModel();
             //特征提取（提取分数最高人脸特征）,适用于单人脸场景
             R<float[]> featureResult1 = faceRecModel.extractTopFaceFeature("src/main/resources/iu_1.jpg");
             if(featureResult1.isSuccess()){
@@ -220,7 +225,8 @@ public class FaceRecDemo {
      */
     @Test
     public void searchFace(){
-        try (FaceRecModel faceRecModel = getFaceRecModelWithDbConfig()){
+        try {
+            FaceRecModel faceRecModel = getFaceRecModelWithDbConfig();
             //等待加载人脸库结束
             while (!faceRecModel.isLoadFaceCompleted()){
                 Thread.sleep(100);
@@ -296,7 +302,8 @@ public class FaceRecDemo {
      */
     @Test
     public void searchFace2(){
-        try (FaceRecModel faceRecModel = getFaceRecModelWithSQLiteConfig()){
+        try {
+            FaceRecModel faceRecModel = getFaceRecModelWithSQLiteConfig();
             //等待加载人脸库结束
             while (!faceRecModel.isLoadFaceCompleted()){
                 Thread.sleep(100);
@@ -367,7 +374,8 @@ public class FaceRecDemo {
     @Test
     public void getFaceInfo(){
         //使用ID获取人脸信息
-        try (FaceRecModel faceRecModel = getFaceRecModelWithSQLiteConfig()){
+        try {
+            FaceRecModel faceRecModel = getFaceRecModelWithSQLiteConfig();
             //等待加载人脸库结束
             while (!faceRecModel.isLoadFaceCompleted()){
                 Thread.sleep(100);
@@ -390,7 +398,8 @@ public class FaceRecDemo {
     @Test
     public void listFaces(){
         //使用ID获取人脸信息
-        try (FaceRecModel faceRecModel = getFaceRecModelWithDbConfig()){
+        try {
+            FaceRecModel faceRecModel = getFaceRecModelWithDbConfig();
             //等待加载人脸库结束
             while (!faceRecModel.isLoadFaceCompleted()){
                 Thread.sleep(100);

examples/face-example/src/main/java/smartai/examples/face/liveness/LivenessDetDemo.java

@@ -133,7 +133,8 @@ public class LivenessDetDemo {
      */
     @Test
     public void testLivenessDetect(){
-        try (LivenessDetModel livenessDetModel = getLivenessDetModel()){
+        try {
+            LivenessDetModel livenessDetModel = getLivenessDetModel();
             R<DetectionResponse> response = livenessDetModel.detect("src/main/resources/liveness/1.jpg");
             if(response.isSuccess()){
                 for (DetectionInfo detectionInfo : response.getData().getDetectionInfoList()){
@@ -152,7 +153,8 @@ public class LivenessDetDemo {
      */
     @Test
     public void testLivenessDetectAndDraw(){
-        try (LivenessDetModel livenessDetModel = getLivenessDetModel()){
+        try {
+            LivenessDetModel livenessDetModel = getLivenessDetModel();
             BufferedImage image = ImageIO.read(new File(Paths.get("src/main/resources/liveness/1.jpg").toAbsolutePath().toString()));
             R<DetectionResponse> response = livenessDetModel.detect(image);
             if(response.isSuccess()){
@@ -175,7 +177,8 @@ public class LivenessDetDemo {
      */
     @Test
     public void testLivenessDetect2(){
-        try (LivenessDetModel livenessDetModel = getLivenessDetModel()){
+        try {
+            LivenessDetModel livenessDetModel = getLivenessDetModel();
             //指定文件夹路径
             File dir = new File("face-example/src/main/resources/liveness");
             File[] files = dir.listFiles();
@@ -198,8 +201,9 @@ public class LivenessDetDemo {
      */
     @Test
     public void testLivenessDetect3(){
-        try (FaceDetModel faceDetectModel = getFaceDetModel();
-             LivenessDetModel livenessDetModel = getLivenessDetModel()){
+        try {
+            FaceDetModel faceDetectModel = getFaceDetModel();
+            LivenessDetModel livenessDetModel = getLivenessDetModel();
             // 将图片路径转换为 BufferedImage
             BufferedImage image = ImageIO.read(new File(Paths.get("src/main/resources/liveness/1.jpg").toAbsolutePath().toString()));
             //人脸检测
@@ -230,8 +234,9 @@ public class LivenessDetDemo {
      */
     @Test
     public void testLivenessDetect4(){
-        try (FaceDetModel faceDetModel = getFaceDetModel();
-             LivenessDetModel livenessDetModel = getMiniVisionLivenessDetModel()){
+        try {
+            FaceDetModel faceDetModel = getFaceDetModel();
+            LivenessDetModel livenessDetModel = getMiniVisionLivenessDetModel();
             // 将图片路径转换为 BufferedImage
             BufferedImage image = ImageIO.read(new File(Paths.get("src/main/resources/liveness/1.jpg").toAbsolutePath().toString()));
             R<DetectionResponse> detResult = faceDetModel.detect(image);
@@ -259,7 +264,8 @@ public class LivenessDetDemo {
      */
     @Test
     public void testLivenessDetectVideo(){
-        try (LivenessDetModel livenessDetModel = getLivenessDetModel()){
+        try {
+            LivenessDetModel livenessDetModel = getLivenessDetModel();
             //视频路径
             R<LivenessResult> livenessStatus = livenessDetModel.detectVideo("video.mp4");
             if (livenessStatus.isSuccess()){
@@ -278,7 +284,8 @@ public class LivenessDetDemo {
      */
     @Test
     public void testLivenessDetectCamera(){
-        try (LivenessDetModel livenessDetModel = getLivenessDetModel()){
+        try {
+            LivenessDetModel livenessDetModel = getLivenessDetModel();
             OpenCV.loadShared();
             VideoCapture capture = new VideoCapture(0);
             if (!capture.isOpened()) {

examples/face-example/src/main/java/smartai/examples/face/quality/FaceQualityDetDemo.java

@@ -78,8 +78,9 @@ public class FaceQualityDetDemo {
      */
     @Test
     public void evaluateBrightness(){
-        try (FaceQualityModel faceQualityModel = getFaceQualityModel();
-                FaceDetModel faceDetModel = getFaceDetModel()){
+        try {
+            FaceQualityModel faceQualityModel = getFaceQualityModel();
+            FaceDetModel faceDetModel = getFaceDetModel();
             //人脸检测
             BufferedImage image = ImageIO.read(new File(Paths.get("src/main/resources/iu_1.jpg").toAbsolutePath().toString()));
             R<DetectionResponse> detectionResponse = faceDetModel.detect(image);
@@ -110,8 +111,9 @@ public class FaceQualityDetDemo {
      */
     @Test
     public void evaluateCompleteness(){
-        try (FaceQualityModel faceQualityModel = getFaceQualityModel();
-             FaceDetModel faceDetModel = getFaceDetModel()){
+        try {
+            FaceQualityModel faceQualityModel = getFaceQualityModel();
+            FaceDetModel faceDetModel = getFaceDetModel();
             //人脸检测
             BufferedImage image = ImageIO.read(new File(Paths.get("src/main/resources/iu_1.jpg").toAbsolutePath().toString()));
             R<DetectionResponse> detectionResponse = faceDetModel.detect(image);
@@ -142,8 +144,9 @@ public class FaceQualityDetDemo {
      */
     @Test
     public void evaluateClarity(){
-        try (FaceQualityModel faceQualityModel = getFaceQualityModel();
-             FaceDetModel faceDetModel = getFaceDetModel()){
+        try {
+            FaceQualityModel faceQualityModel = getFaceQualityModel();
+            FaceDetModel faceDetModel = getFaceDetModel();
             //人脸检测
             BufferedImage image = ImageIO.read(new File(Paths.get("src/main/resources/iu_1.jpg").toAbsolutePath().toString()));
             R<DetectionResponse> detectionResponse = faceDetModel.detect(image);
@@ -174,8 +177,9 @@ public class FaceQualityDetDemo {
      */
     @Test
     public void evaluatePose(){
-        try (FaceQualityModel faceQualityModel = getFaceQualityModel();
-             FaceDetModel faceDetModel = getFaceDetModel()){
+        try {
+            FaceQualityModel faceQualityModel = getFaceQualityModel();
+            FaceDetModel faceDetModel = getFaceDetModel();
             //人脸检测
             BufferedImage image = ImageIO.read(new File(Paths.get("src/main/resources/iu_1.jpg").toAbsolutePath().toString()));
             R<DetectionResponse> detectionResponse = faceDetModel.detect(image);
@@ -207,8 +211,9 @@ public class FaceQualityDetDemo {
      */
     @Test
     public void evaluateResolution(){
-        try (FaceQualityModel faceQualityModel = getFaceQualityModel();
-             FaceDetModel faceDetModel = getFaceDetModel()){
+        try {
+            FaceQualityModel faceQualityModel = getFaceQualityModel();
+            FaceDetModel faceDetModel = getFaceDetModel();
             //人脸检测
             BufferedImage image = ImageIO.read(new File(Paths.get("src/main/resources/iu_1.jpg").toAbsolutePath().toString()));
             R<DetectionResponse> detectionResponse = faceDetModel.detect(image);
@@ -240,8 +245,9 @@ public class FaceQualityDetDemo {
      */
     @Test
     public void evaluateAll(){
-        try (FaceQualityModel faceQualityModel = getFaceQualityModel();
-             FaceDetModel faceDetModel = getFaceDetModel()){
+        try {
+            FaceQualityModel faceQualityModel = getFaceQualityModel();
+            FaceDetModel faceDetModel = getFaceDetModel();
             //人脸检测
             BufferedImage image = ImageIO.read(new File(Paths.get("src/main/resources/iu_1.jpg").toAbsolutePath().toString()));
             R<DetectionResponse> detectionResponse = faceDetModel.detect(image);

examples/objectdetection-example/pom.xml

@@ -12,7 +12,7 @@
         <maven.compiler.source>11</maven.compiler.source>
         <maven.compiler.target>11</maven.compiler.target>
         <project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
-        <smartjavaai.version>1.0.20</smartjavaai.version>
+        <smartjavaai.version>1.0.22</smartjavaai.version>
         <!--如果打包运行，需要替换成你的main-->
         <exec.mainClass>smartai.examples.objectdetection.ObjectDetection</exec.mainClass>
 

examples/objectdetection-example/src/main/java/smartai/examples/objectdetection/ObjectDetection.java

@@ -41,10 +41,8 @@ import java.awt.image.BufferedImage;
 import java.io.File;
 import java.io.IOException;
 import java.nio.file.Paths;
-import java.util.ArrayList;
-import java.util.Iterator;
+import java.util.*;
 import java.util.List;
-import java.util.Objects;
 import java.util.concurrent.Callable;
 import java.util.concurrent.ExecutorService;
 import java.util.concurrent.Executors;
@@ -71,7 +69,8 @@ public class ObjectDetection {
     @Test
     public void objectDetection(){
         //默认cpu
-        try (DetectorModel detectorModel = ObjectDetectionModelFactory.getInstance().getModel()){
+        try {
+            DetectorModel detectorModel = ObjectDetectionModelFactory.getInstance().getModel();
             DetectionResponse detectionResponse = detectorModel.detect("src/main/resources/object_detection.jpg");
             log.info("目标检测结果：{}", JSONObject.toJSONString(detectionResponse));
         } catch (Exception e) {
@@ -84,10 +83,15 @@ public class ObjectDetection {
      */
     @Test
     public void objectDetection2(){
-        DetectorModelConfig config = new DetectorModelConfig();
-        config.setModelEnum(DetectorModelEnum.SSD_300_RESNET50);//检测模型，目前支持19种预置模型
-        config.setDevice(device);
-        try (DetectorModel detectorModel = ObjectDetectionModelFactory.getInstance().getModel(config)){
+        try {
+            DetectorModelConfig config = new DetectorModelConfig();
+            config.setModelEnum(DetectorModelEnum.SSD_300_RESNET50);//检测模型，目前支持19种预置模型
+            // 指定允许的类别
+//            config.setAllowedClasses(Arrays.asList("person"));
+            //指定返回检测数量
+            config.setTopK(100);
+            config.setDevice(device);
+            DetectorModel detectorModel = ObjectDetectionModelFactory.getInstance().getModel(config);
             DetectionResponse detectionResponse = detectorModel.detect("src/main/resources/dog_bike_car.jpg");
             log.info("目标检测结果：{}", JSONObject.toJSONString(detectionResponse));
         } catch (Exception e) {
@@ -100,7 +104,8 @@ public class ObjectDetection {
      */
     @Test
     public void objectDetectionAndDraw(){
-        try (DetectorModel detectorModel = ObjectDetectionModelFactory.getInstance().getModel()){
+        try {
+            DetectorModel detectorModel = ObjectDetectionModelFactory.getInstance().getModel();
             detectorModel.detectAndDraw("src/main/resources/object_detection.jpg","output/object_detection_detected.png");
         } catch (Exception e) {
             e.printStackTrace();
@@ -112,7 +117,8 @@ public class ObjectDetection {
      */
     @Test
     public void objectDetectionAndDraw2(){
-        try (DetectorModel detectorModel = ObjectDetectionModelFactory.getInstance().getModel()){
+        try {
+            DetectorModel detectorModel = ObjectDetectionModelFactory.getInstance().getModel();
             String imagePath = "src/main/resources/object_detection.jpg";
             BufferedImage image = ImageIO.read(new File(Paths.get(imagePath).toAbsolutePath().toString()));
             //可以根据后续业务场景使用detectedImage
@@ -131,15 +137,16 @@ public class ObjectDetection {
      */
     @Test
     public void objectDetectionWithOfficialModel(){
-        DetectorModelConfig config = new DetectorModelConfig();
-        config.setThreshold(0.3f);
-        //也支持YoloV8：YOLOV8_OFFICIAL 模型可以从文档中提供的地址下载
-        config.setModelEnum(DetectorModelEnum.YOLOV12_OFFICIAL);//检测模型，目前支持19种模型
-        // 指定模型路径，需要更改为自己的模型路径
-        config.setModelPath("/Users/xxx/Documents/yolov12n.onnx");
-        config.setDevice(device);
-        //一定要将yolo官方的类别文件：synset.txt（文档中下载）放在模型同目录下，否则报错
-        try (DetectorModel detectorModel = ObjectDetectionModelFactory.getInstance().getModel(config)){
+        try {
+            DetectorModelConfig config = new DetectorModelConfig();
+            config.setThreshold(0.3f);
+            //也支持YoloV8：YOLOV8_OFFICIAL 模型可以从文档中提供的地址下载
+            config.setModelEnum(DetectorModelEnum.YOLOV12_OFFICIAL);//检测模型，目前支持19种模型
+            // 指定模型路径，需要更改为自己的模型路径
+            config.setModelPath("/Users/xxx/Documents/yolov12n.onnx");
+            config.setDevice(device);
+            //一定要将yolo官方的类别文件：synset.txt（文档中下载）放在模型同目录下，否则报错
+            DetectorModel detectorModel = ObjectDetectionModelFactory.getInstance().getModel(config);
             DetectionResponse detect = detectorModel.detect("src/main/resources/dog_bike_car.jpg");
             log.info("目标检测结果：{}", JSONObject.toJSONString(detect));
         } catch (Exception e) {
@@ -152,16 +159,21 @@ public class ObjectDetection {
      */
     @Test
     public void objectDetectionWithCustomModel(){
-        DetectorModelConfig config = new DetectorModelConfig();
-        //也支持YoloV8：YOLOV8_CUSTOM 模型需要自己训练，训练教程可以查看文档
-        config.setModelEnum(DetectorModelEnum.YOLOV12_CUSTOM);//自定义YOLOV12模型
-        // 指定模型路径，需要更改为自己的模型路径
-        config.setModelPath("/Users/xxx/Documents/develop/fire_model/best.onnx");
-        config.putCustomParam("width", 640);//resize 宽
-        config.putCustomParam("height", 640);// resize 高
-        config.putCustomParam("nmsThreshold", 0.5f);
-        config.setDevice(device);
-        try (DetectorModel detectorModel = ObjectDetectionModelFactory.getInstance().getModel(config)){
+        try {
+            DetectorModelConfig config = new DetectorModelConfig();
+            //也支持YoloV8：YOLOV8_CUSTOM 模型需要自己训练，训练教程可以查看文档
+            config.setModelEnum(DetectorModelEnum.YOLOV12_CUSTOM);//自定义YOLOV12模型
+            // 指定模型路径，需要更改为自己的模型路径
+            config.setModelPath("/Users/xxx/Documents/develop/fire_model/best.onnx");
+            config.putCustomParam("width", 640);//resize 宽
+            config.putCustomParam("height", 640);// resize 高
+            config.putCustomParam("nmsThreshold", 0.5f);
+            // 指定允许的类别
+//            config.setAllowedClasses(Arrays.asList("person"));
+            //指定返回检测数量
+            config.setTopK(100);
+            config.setDevice(device);
+            DetectorModel detectorModel = ObjectDetectionModelFactory.getInstance().getModel(config);
             DetectionResponse detect = detectorModel.detect("src/main/resources/dog_bike_car.jpg");
             log.info("目标检测结果：{}", JSONObject.toJSONString(detect));
         } catch (Exception e) {
@@ -176,7 +188,8 @@ public class ObjectDetection {
      */
     @Test
     public void testDetectCamera(){
-        try (DetectorModel detectorModel = ObjectDetectionModelFactory.getInstance().getModel()){
+        try {
+            DetectorModel detectorModel = ObjectDetectionModelFactory.getInstance().getModel();
             OpenCV.loadShared();
             VideoCapture capture = new VideoCapture(0);
             if (!capture.isOpened()) {

examples/ocr-examples/pom.xml

@@ -12,7 +12,7 @@
         <maven.compiler.source>11</maven.compiler.source>
         <maven.compiler.target>11</maven.compiler.target>
         <project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
-        <smartjavaai.version>1.0.20</smartjavaai.version>
+        <smartjavaai.version>1.0.22</smartjavaai.version>
         <!--如果打包运行，需要替换成你的main-->
         <exec.mainClass>smartai.examples.ocr.common.OcrRecognizeDemo</exec.mainClass>
 

examples/ocr-examples/src/main/java/smartai/examples/ocr/common/OcrDetectionDemo.java

@@ -11,7 +11,6 @@ import cn.smartjavaai.ocr.entity.OcrInfo;
 import cn.smartjavaai.ocr.enums.CommonDetModelEnum;
 import cn.smartjavaai.ocr.factory.OcrModelFactory;
 import cn.smartjavaai.ocr.model.common.detect.OcrCommonDetModel;
-import cn.smartjavaai.ocr.opencv.OcrOpenCVUtils;
 import cn.smartjavaai.ocr.utils.OcrUtils;
 import com.alibaba.fastjson.JSONObject;
 import lombok.extern.slf4j.Slf4j;
@@ -60,7 +59,8 @@ public class OcrDetectionDemo {
      */
     @Test
     public void detect(){
-        try (OcrCommonDetModel model = getDetectionModel()){
+        try {
+            OcrCommonDetModel model = getDetectionModel();
             List<OcrBox> boxes = model.detect("src/main/resources/ocr_1.jpg");
             log.info("OCR检测结果：{}", JSONObject.toJSONString(boxes));
         } catch (Exception e) {
@@ -77,7 +77,8 @@ public class OcrDetectionDemo {
      */
     @Test
     public void detectAndDraw(){
-        try (OcrCommonDetModel model = getDetectionModel()){
+        try {
+            OcrCommonDetModel model = getDetectionModel();
             model.detectAndDraw("src/main/resources/ocr_1.jpg",  "output/ocr_1_detected.jpg");
         } catch (Exception e) {
             e.printStackTrace();
@@ -94,7 +95,8 @@ public class OcrDetectionDemo {
      */
     @Test
     public void batchDetect(){
-        try (OcrCommonDetModel model = getDetectionModel()){
+        try {
+            OcrCommonDetModel model = getDetectionModel();
             //批量检测要求图片宽高一致
             String folderPath = "/Users/xxx/Downloads/testing33";
             //读取文件夹中所有图片

examples/ocr-examples/src/main/java/smartai/examples/ocr/common/OcrDirectionDetDemo.java

@@ -67,7 +67,8 @@ public class OcrDirectionDetDemo {
      */
     @Test
     public void detect(){
-        try (OcrDirectionModel directionModel = getDirectionModel()){
+        try {
+            OcrDirectionModel directionModel = getDirectionModel();
             List<OcrItem> itemList = directionModel.detect("src/main/resources/ocr_1.jpg");
             log.info("OCR方向检测结果1：{}", JSONObject.toJSONString(itemList));
         } catch (Exception e) {
@@ -84,7 +85,8 @@ public class OcrDirectionDetDemo {
      */
     @Test
     public void detectAndDraw(){
-        try (OcrDirectionModel directionModel = getDirectionModel()){
+        try {
+            OcrDirectionModel directionModel = getDirectionModel();
             directionModel.detectAndDraw("src/main/resources/ocr_3.jpg",  "output/ocr_3_detected.png");
         } catch (Exception e) {
             e.printStackTrace();

examples/ocr-examples/src/main/java/smartai/examples/ocr/common/OcrRecognizeDemo.java

@@ -106,7 +106,8 @@ public class OcrRecognizeDemo {
      */
     @Test
     public void recognize(){
-        try (OcrCommonRecModel recModel = getRecModel()){
+        try {
+            OcrCommonRecModel recModel = getRecModel();
             //不带方向矫正，分行返回文本
             OcrRecOptions options = new OcrRecOptions(false, true);
             OcrInfo ocrInfo = recModel.recognize("src/main/resources/ocr_2.jpg",options);
@@ -127,7 +128,8 @@ public class OcrRecognizeDemo {
      */
     @Test
     public void recognizeHandWriting(){
-        try (OcrCommonRecModel recModel = getRecModel()){
+        try {
+            OcrCommonRecModel recModel = getRecModel();
             OcrInfo ocrInfo = recModel.recognize("src/main/resources/handwriting_1.jpg",new OcrRecOptions());
             log.info("OCR识别结果：{}", JSONObject.toJSONString(ocrInfo));
         } catch (Exception e) {
@@ -146,7 +148,8 @@ public class OcrRecognizeDemo {
      */
     @Test
     public void recognize2(){
-        try (OcrCommonRecModel recModel = getRecModelWithDirection()){
+        try {
+            OcrCommonRecModel recModel = getRecModelWithDirection();
             //带方向矫正，分行返回文本
             OcrRecOptions options = new OcrRecOptions(true, true);
             OcrInfo ocrInfo = recModel.recognize("src/main/resources/ocr_3.jpg",options);
@@ -168,7 +171,8 @@ public class OcrRecognizeDemo {
      */
     @Test
     public void recognizeAndDraw(){
-        try (OcrCommonRecModel recModel = getRecModelWithDirection()){
+        try {
+            OcrCommonRecModel recModel = getRecModelWithDirection();
             int fontSize = 18;
             recModel.recognizeAndDraw("src/main/resources/general_ocr_002.png", "output/ocr_4_recognized.jpg", fontSize, new OcrRecOptions());
         } catch (Exception e) {
@@ -184,7 +188,8 @@ public class OcrRecognizeDemo {
      */
     @Test
     public void batchRecognize(){
-        try (OcrCommonRecModel recModel = getRecModelWithDirection()){
+        try {
+            OcrCommonRecModel recModel = getRecModelWithDirection();
             //批量检测要求图片宽高一致
             String folderPath = "/Users/xxx/Downloads/testing33";
             //读取文件夹中所有图片

examples/ocr-examples/src/main/java/smartai/examples/ocr/plate/PlateRecDemo.java

@@ -0,0 +1,81 @@
+package smartai.examples.ocr.plate;
+
+import ai.djl.util.JsonUtils;
+import cn.smartjavaai.common.entity.R;
+import cn.smartjavaai.common.enums.DeviceEnum;
+import cn.smartjavaai.common.utils.ImageUtils;
+import cn.smartjavaai.ocr.config.PlateDetModelConfig;
+import cn.smartjavaai.ocr.config.PlateRecModelConfig;
+import cn.smartjavaai.ocr.entity.PlateInfo;
+import cn.smartjavaai.ocr.enums.PlateDetModelEnum;
+import cn.smartjavaai.ocr.enums.PlateRecModelEnum;
+import cn.smartjavaai.ocr.factory.PlateModelFactory;
+import cn.smartjavaai.ocr.model.plate.PlateDetModel;
+import cn.smartjavaai.ocr.model.plate.PlateRecModel;
+import lombok.extern.slf4j.Slf4j;
+import org.junit.Test;
+
+import java.io.File;
+import java.util.List;
+
+/**
+ * @author dwj
+ */
+@Slf4j
+public class PlateRecDemo {
+
+    //设备类型
+    public static DeviceEnum device = DeviceEnum.CPU;
+
+    /**
+     * 获取车牌检测模型
+     * @return
+     */
+    public PlateDetModel getPlateDetModel() {
+        PlateDetModelConfig config = new PlateDetModelConfig();
+        config.setModelEnum(PlateDetModelEnum.YOLOV5);
+        config.setModelPath("/Users/wenjie/Downloads/modelscope-2b3747db53adb48370c6edaccac56eb1ca0da1b5/onnx_export/model.onnx");
+//        config.setModelPath("/Users/wenjie/Documents/develop/model/plate/yolov8s.onnx");
+        config.setModelPath("/Users/wenjie/Downloads/数据集/车牌/onnx/plate_detect.onnx");
+        config.setPredictorPoolSize(3);
+        config.setDevice(device);
+        return PlateModelFactory.getInstance().getDetModel(config);
+    }
+
+    /**
+     * 获取车牌识别模型
+     * @return
+     */
+    public PlateRecModel getPlateRecModel() {
+        PlateRecModelConfig recModelConfig = new PlateRecModelConfig();
+        recModelConfig.setModelEnum(PlateRecModelEnum.PLATE_REC_CRNN);
+        recModelConfig.setModelPath("/Users/wenjie/Downloads/数据集/车牌/onnx/plate_rec_color.onnx");
+        recModelConfig.setPlateDetModel(getPlateDetModel());
+        return PlateModelFactory.getInstance().getRecModel(recModelConfig);
+    }
+
+    @Test
+    public void testDetect() {
+        PlateRecModel plateRecModel = getPlateRecModel();
+        R<List<PlateInfo>> result = plateRecModel.recognize("src/main/resources/plate/Quicker_20220930_180856.png");
+        if(result.isSuccess()){
+            log.info("车牌识别结果：{}", JsonUtils.toJson(result.getData()));
+        }else{
+            log.error("车牌识别失败：{}", result.getMessage());
+        }
+    }
+
+    @Test
+    public void recognizeAndDraw() {
+        PlateRecModel plateRecModel = getPlateRecModel();
+        R<Void> result = plateRecModel.recognizeAndDraw("src/main/resources/plate/single_green.jpg", "output/plate_recognized2.jpg");
+        if(result.isSuccess()){
+            log.info("车牌识别成功");
+        }else{
+            log.error("车牌识别失败：{}", result.getMessage());
+        }
+
+    }
+
+
+}

examples/ocr-examples/src/main/java/smartai/examples/ocr/table/TableRecDemo.java

@@ -113,10 +113,11 @@ public class TableRecDemo {
      */
     @Test
     public void recognize(){
-        try (TableStructureModel tableStructureModel = getTableStructureModel();
-             OcrCommonDetModel detModel = getDetectionModel();
-             OcrCommonRecModel recModel = getRecModel();
-             OcrDirectionModel directionModel = getDirectionModel()){
+        try {
+            TableStructureModel tableStructureModel = getTableStructureModel();
+            OcrCommonDetModel detModel = getDetectionModel();
+            OcrCommonRecModel recModel = getRecModel();
+            OcrDirectionModel directionModel = getDirectionModel();
             //创建表格识别器
             TableRecognizer tableRecognizer = TableRecognizer.builder()
                     .withStructureModel(tableStructureModel)

examples/translation-example/pom.xml

@@ -12,7 +12,7 @@
         <maven.compiler.source>11</maven.compiler.source>
         <maven.compiler.target>11</maven.compiler.target>
         <project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
-        <smartjavaai.version>1.0.20</smartjavaai.version>
+        <smartjavaai.version>1.0.22</smartjavaai.version>
         <!--如果打包运行，需要替换成你的main-->
         <exec.mainClass>smartai.examples.nlp.translation.TranslationDemo</exec.mainClass>
 

pom.xml

@@ -7,7 +7,7 @@
     <name>SmartJavaAI</name>
     <groupId>cn.smartjavaai</groupId>
     <artifactId>smartjavaai-parent</artifactId>
-    <version>1.0.20</version>
+    <version>1.0.22</version>
     <packaging>pom</packaging>
     <description>SmartJavaAI</description>
     <modules>

smartjavaai-all/pom.xml

@@ -6,11 +6,11 @@
     <parent>
         <groupId>cn.smartjavaai</groupId>
         <artifactId>smartjavaai-parent</artifactId>
-        <version>1.0.20</version>
+        <version>1.0.22</version>
     </parent>
 
     <artifactId>smartjavaai-all</artifactId>
-    <version>1.0.20</version>
+    <version>1.0.22</version>
     <name>${project.artifactId}</name>
     <description>SmartJavaAI</description>
     <url>https://github.com/geekwenjie/SmartJavaAI</url>

smartjavaai-bom/pom.xml

@@ -6,10 +6,10 @@
     <parent>
         <groupId>cn.smartjavaai</groupId>
         <artifactId>smartjavaai-parent</artifactId>
-        <version>1.0.20</version>
+        <version>1.0.22</version>
     </parent>
 
-    <version>1.0.20</version>
+    <version>1.0.22</version>
     <artifactId>smartjavaai-bom</artifactId>
     <name>smartjavaai-bom</name>
     <description>统一版本管理的 BOM 包，同时支持 import 和全量依赖</description>

smartjavaai-common/pom.xml

@@ -6,7 +6,7 @@
     <parent>
         <groupId>cn.smartjavaai</groupId>
         <artifactId>smartjavaai-parent</artifactId>
-        <version>1.0.20</version>
+        <version>1.0.22</version>
     </parent>
 
     <name>smartjavaai-common</name>

smartjavaai-common/src/main/java/cn/smartjavaai/common/config/ModelConfig.java

@@ -29,6 +29,11 @@ public class ModelConfig {
      */
     private String batchifier;
 
+    /**
+     * 模型预测器池大小(默认为cpu核心数)
+     */
+    private int predictorPoolSize;
+
     /**
      * 个性化配置（按模型类型动态解析）
      */

smartjavaai-common/src/main/java/cn/smartjavaai/common/entity/R.java

@@ -57,6 +57,7 @@ public class R<T> {
         NO_FACE_DETECTED(3, "未检测到人脸"),
         PARAM_ERROR(4, "参数错误"),
         INVALID_VIDEO(5, "视频无效"),
+        NO_OBJECT_DETECTED(6, "未检测到目标"),
         Unknown(-1,  "未知错误");
 
         private final int code;

smartjavaai-common/src/main/java/cn/smartjavaai/common/pool/PredictorFactory.java

@@ -3,6 +3,7 @@ package cn.smartjavaai.common.pool;
 import ai.djl.inference.Predictor;
 import ai.djl.repository.zoo.ZooModel;
 import ai.djl.translate.Translator;
+import lombok.extern.slf4j.Slf4j;
 import org.apache.commons.pool2.BasePooledObjectFactory;
 import org.apache.commons.pool2.PooledObject;
 import org.apache.commons.pool2.impl.DefaultPooledObject;
@@ -12,6 +13,7 @@ import org.apache.commons.pool2.impl.DefaultPooledObject;
  * @author dwj
  * @date 2025/4/8
  */
+@Slf4j
 public class PredictorFactory<I, O> extends BasePooledObjectFactory<Predictor<I, O>> {
     private final ZooModel<I, O> model;
 
@@ -21,6 +23,7 @@ public class PredictorFactory<I, O> extends BasePooledObjectFactory<Predictor<I,
 
     @Override
     public Predictor<I, O> create() {
+        log.debug("create predictor");
         return model.newPredictor();
     }
 
@@ -31,6 +34,7 @@ public class PredictorFactory<I, O> extends BasePooledObjectFactory<Predictor<I,
 
     @Override
     public void destroyObject(PooledObject<Predictor<I, O>> p) {
+        log.debug("close predictor");
         p.getObject().close();
     }
 }

smartjavaai-common/src/main/java/cn/smartjavaai/common/utils/ImageUtils.java

@@ -8,6 +8,9 @@ import ai.djl.ndarray.NDArray;
 import cn.smartjavaai.common.entity.DetectionRectangle;
 import cn.smartjavaai.common.entity.DetectionResponse;
 import org.opencv.core.Mat;
+import org.opencv.core.Scalar;
+import org.opencv.core.Size;
+import org.opencv.imgproc.Imgproc;
 
 import javax.imageio.ImageIO;
 import java.awt.*;
@@ -364,6 +367,7 @@ public class ImageUtils {
         int width = metrics.stringWidth(text) + padding * 2 - stroke / 2;
         int height = metrics.getHeight() + metrics.getDescent();
         int ascent = metrics.getAscent();
+        y = Math.max(0, y - height);
         java.awt.Rectangle background = new java.awt.Rectangle(x, y, width, height);
         g.fill(background);
         g.setPaint(Color.WHITE);
@@ -468,6 +472,33 @@ public class ImageUtils {
         return true;
     }
 
+    /**
+     * 在图像上绘制带白色背景、黑色文字的文本
+     */
+    public static void putTextWithBackground(Mat image, String text, org.opencv.core.Point origin, Scalar textColor, Scalar backgroundColor, int padding) {
+        // 默认字体
+        int font = Imgproc.FONT_HERSHEY_SCRIPT_SIMPLEX;
+        // 默认字体缩放大小
+        double fontScale = 1.0;
+        //线条粗细
+        int thickness = 2;
+        //获取文字大小
+        int[] baseLine = new int[1];
+        Size textSize = Imgproc.getTextSize(text, font, fontScale, thickness, baseLine);
+        int textWidth = (int) textSize.width;
+        int textHeight = (int) textSize.height;
+
+        //计算带padding的背景框
+        org.opencv.core.Point bgTopLeft = new org.opencv.core.Point(origin.x - padding, origin.y - textHeight - padding);
+        org.opencv.core.Point bgBottomRight = new org.opencv.core.Point(origin.x + textWidth + padding, origin.y + baseLine[0] + padding);
+
+        //绘制背景矩形
+        Imgproc.rectangle(image, bgTopLeft, bgBottomRight, backgroundColor, Imgproc.FILLED);
+
+        //绘制文字（黑色）
+        Imgproc.putText(image, text, origin, font, fontScale, textColor, thickness);
+    }
+
 
 
 }

smartjavaai-common/src/main/java/cn/smartjavaai/common/utils/LetterBoxUtils.java

@@ -0,0 +1,124 @@
+package cn.smartjavaai.common.utils;
+
+import ai.djl.modality.cv.util.NDImageUtils;
+import ai.djl.ndarray.NDArray;
+import ai.djl.ndarray.NDManager;
+import ai.djl.ndarray.index.NDIndex;
+import ai.djl.ndarray.types.DataType;
+import ai.djl.ndarray.types.Shape;
+
+import java.util.Arrays;
+
+/**
+ * 按比例缩放，剩余空间用指定颜色填充
+ * @author dwj
+ */
+public class LetterBoxUtils {
+
+    public enum PaddingPosition {
+        CENTER, LEFT_TOP, RIGHT_BOTTOM
+    }
+
+    public static class ResizeResult {
+        public NDArray image;
+        public float r;
+        public int left;
+        public int top;
+    }
+
+    public static ResizeResult letterboxWithMeta(NDArray paddingImg, float r, int left, int top) {
+        // ... letterbox 逻辑不变
+        ResizeResult result = new ResizeResult();
+        result.image = paddingImg;
+        result.r = r;
+        result.left = left;
+        result.top = top;
+        return result;
+    }
+
+
+    /**
+     * 按比例缩放 + padding
+     *
+     * @param img           原图 NDArray HWC
+     * @param targetW       目标宽度
+     * @param targetH       目标高度
+     * @param padColor      padding 填充颜色，RGB 归一化 0-1
+     * @param position      padding 位置：CENTER / LEFT_TOP / RIGHT_BOTTOM
+     * @return              处理后的 NDArray
+     */
+    public static ResizeResult letterbox(NDManager manager, NDArray img, int targetW, int targetH, float padColor, PaddingPosition position) {
+        long origH = img.getShape().get(0);
+        long origW = img.getShape().get(1);
+
+        float r = Math.min(targetW / (float) origW, targetH / (float) origH);
+        int newW = Math.round(origW * r);
+        int newH = Math.round(origH * r);
+
+        img = NDImageUtils.resize(img, newW, newH);  // HWC 0~1
+
+//        NDArray paddingImg = manager
+//                .full(new Shape(targetW, targetH, 3), padColor, DataType.UINT8);
+
+        NDArray paddingImg = manager.zeros(new Shape(targetW, targetH, 3), DataType.FLOAT32);
+        paddingImg = paddingImg.add(114);
+
+        int padW = targetW - newW;
+        int padH = targetH - newH;
+        int top = 0, left = 0;
+
+        switch (position) {
+            case CENTER:
+                left = padW / 2;
+                top = padH / 2;
+                break;
+            case LEFT_TOP:
+                left = 0;
+                top = 0;
+                break;
+            case RIGHT_BOTTOM:
+                left = padW;
+                top = padH;
+                break;
+        }
+        paddingImg.set(new NDIndex(String.format("%d:%d,%d:%d", top, top + newH, left, left + newW)), img);
+        return letterboxWithMeta(paddingImg, r, left, top);
+    }
+
+
+    /**
+     * 恢复缩放后的 box
+     * @param boxes
+     * @param scaleRatio
+     * @param left
+     * @param top
+     * @param keypointStart
+     * @param keypointDim
+     * @return
+     */
+    public static NDArray restoreBox(NDArray boxes, float scaleRatio, float left, float top, int keypointStart, int keypointDim) {
+        // 处理 bbox
+        NDArray x1 = boxes.get(":, 0").sub(left).div(scaleRatio);
+        NDArray y1 = boxes.get(":, 1").sub(top).div(scaleRatio);
+        NDArray x2 = boxes.get(":, 2").sub(left).div(scaleRatio);
+        NDArray y2 = boxes.get(":, 3").sub(top).div(scaleRatio);
+
+        boxes.set(new NDIndex(":, 0"), x1);
+        boxes.set(new NDIndex(":, 1"), y1);
+        boxes.set(new NDIndex(":, 2"), x2);
+        boxes.set(new NDIndex(":, 3"), y2);
+
+        if (keypointDim > 0) {
+            for (int i = 0; i < keypointDim; i += 2) {
+                int xIdx = keypointStart + i;
+                int yIdx = keypointStart + i + 1;
+                NDArray keyX = boxes.get(":, " + xIdx).sub(left).div(scaleRatio);
+                NDArray keyY = boxes.get(":, " + yIdx).sub(top).div(scaleRatio);
+                boxes.set(new NDIndex(":, " + xIdx), keyX);
+                boxes.set(new NDIndex(":, " + yIdx), keyY);
+            }
+        }
+        return boxes;
+    }
+
+}

smartjavaai-common/src/main/java/cn/smartjavaai/common/utils/NMSUtils.java

@@ -0,0 +1,67 @@
+package cn.smartjavaai.common.utils;
+
+import ai.djl.ndarray.NDArray;
+
+import java.util.ArrayList;
+import java.util.List;
+
+/**
+ * @author dwj
+ * @date 2025/7/23
+ */
+public class NMSUtils {
+
+    /**
+     * 通用 NMS 方法，输入 NDArray 形式的 boxes 和 scores，返回保留的索引列表
+     *
+     * @param boxes   NDArray 形状为 (N, 4)，格式为 [x1, y1, x2, y2]
+     * @param scores  NDArray 形状为 (N,) 或 (N,1)，每个 box 的置信度
+     * @param iouThreshold IOU 阈值，超过该阈值则认为有重叠
+     * @return 保留框的索引列表
+     */
+    public static int[] nms(NDArray boxes, NDArray scores, float iouThreshold) {
+        if (boxes.isEmpty()) {
+            return new int[0];
+        }
+
+        NDArray x1 = boxes.get(":, 0");
+        NDArray y1 = boxes.get(":, 1");
+        NDArray x2 = boxes.get(":, 2");
+        NDArray y2 = boxes.get(":, 3");
+
+        NDArray areas = x2.sub(x1).add(1).mul(y2.sub(y1).add(1));
+
+        // 按照置信度降序排序
+        NDArray order = scores.argSort().flip(0);
+
+        List<Integer> keep = new ArrayList<>();
+
+        while (order.size() > 0) {
+            int idx = (int)order.getLong(0);
+            keep.add(idx);
+
+            if (order.size() == 1) break;
+
+            NDArray currentBox = boxes.get(idx);
+            NDArray others = boxes.get(order);
+
+            NDArray xx1 = x1.get(order).maximum(x1.get(idx));
+            NDArray yy1 = y1.get(order).maximum(y1.get(idx));
+            NDArray xx2 = x2.get(order).minimum(x2.get(idx));
+            NDArray yy2 = y2.get(order).minimum(y2.get(idx));
+
+            NDArray w = xx2.sub(xx1).add(1).maximum(0);
+            NDArray h = yy2.sub(yy1).add(1).maximum(0);
+            NDArray inter = w.mul(h);
+
+            NDArray remAreas = areas.get(order);
+            NDArray union = remAreas.add(areas.get(idx)).sub(inter);
+            NDArray iou = inter.div(union);
+
+            NDArray mask = iou.lte(iouThreshold);
+            order = order.get(mask);
+        }
+        return keep.stream().mapToInt(i -> i).toArray();
+    }
+
+}

smartjavaai-common/src/main/java/cn/smartjavaai/common/utils/OpenCVUtils.java

@@ -124,4 +124,20 @@ public class OpenCVUtils {
         mat.put(0, 0, data);
         return mat;
     }
+
+    /**
+     * 透视变换
+     *
+     * @param src
+     * @param srcPoints
+     * @param dstPoints
+     * @return
+     */
+    public static Mat perspectiveTransform(Mat src, Mat srcPoints, Mat dstPoints) {
+        Mat dst = src.clone();
+        Mat warp_mat = Imgproc.getPerspectiveTransform(srcPoints, dstPoints);
+        Imgproc.warpPerspective(src, dst, warp_mat, dst.size());
+        warp_mat.release();
+        return dst;
+    }
 }

smartjavaai-common/src/main/java/cn/smartjavaai/common/utils/PointUtils.java

@@ -0,0 +1,70 @@
+package cn.smartjavaai.common.utils;
+
+import ai.djl.modality.cv.output.Point;
+
+import java.util.Arrays;
+import java.util.Comparator;
+import java.util.List;
+
+/**
+ * @author dwj
+ */
+public class PointUtils {
+
+    /**
+     * 对 4 个关键点进行排序，顺序为：
+     * 左上、右上、右下、左下
+     */
+    public static List<Point> orderPoints(List<Point> points) {
+        if (points == null || points.size() != 4) {
+            throw new IllegalArgumentException("必须提供 4 个点");
+        }
+
+        // 按 X 坐标升序排列
+        points.sort(Comparator.comparingDouble(Point::getX));
+
+        List<Point> left = points.subList(0, 2);
+        List<Point> right = points.subList(2, 4);
+
+        // 左侧两点按 Y 排序：上为 tl，下为 bl
+        Point tl = left.get(0).getY() < left.get(1).getY() ? left.get(0) : left.get(1);
+        Point bl = left.get(0).getY() >= left.get(1).getY() ? left.get(0) : left.get(1);
+
+        // 右侧两点按 Y 排序：上为 tr，下为 br
+        Point tr = right.get(0).getY() < right.get(1).getY() ? right.get(0) : right.get(1);
+        Point br = right.get(0).getY() >= right.get(1).getY() ? right.get(0) : right.get(1);
+
+        return Arrays.asList(tl, tr, br, bl);
+    }
+
+    /**
+     * 欧式距离计算
+     *
+     * @param point1
+     * @param point2
+     * @return
+     */
+    public static float distance(float[] point1, float[] point2) {
+        float disX = point1[0] - point2[0];
+        float disY = point1[1] - point2[1];
+        float dis = (float) Math.sqrt(disX * disX + disY * disY);
+        return dis;
+    }
+
+    /**
+     * 欧式距离计算
+     *
+     * @param point1
+     * @param point2
+     * @return
+     */
+    public static float distance(Point point1, Point point2) {
+        double disX = point1.getX() - point2.getX();
+        double disY = point1.getY() - point2.getY();
+        float dis = (float) Math.sqrt(disX * disX + disY * disY);
+        return dis;
+    }
+
+
+
+}

smartjavaai-face/pom.xml

@@ -6,11 +6,11 @@
     <parent>
         <groupId>cn.smartjavaai</groupId>
         <artifactId>smartjavaai-parent</artifactId>
-        <version>1.0.20</version>
+        <version>1.0.22</version>
     </parent>
 
     <artifactId>smartjavaai-face</artifactId>
-    <version>1.0.20</version>
+    <version>1.0.22</version>
     <name>smartjavaai-face</name>
     <description>SmartJavaAI</description>
     <url>https://github.com/geekwenjie/SmartJavaAI</url>

smartjavaai-face/src/main/java/cn/smartjavaai/face/config/FaceAttributeConfig.java

@@ -1,5 +1,6 @@
 package cn.smartjavaai.face.config;
 
+import cn.smartjavaai.common.config.ModelConfig;
 import cn.smartjavaai.common.enums.DeviceEnum;
 import cn.smartjavaai.face.enums.FaceAttributeModelEnum;
 import lombok.Data;
@@ -9,7 +10,7 @@ import lombok.Data;
  * @author dwj
  */
 @Data
-public class FaceAttributeConfig {
+public class FaceAttributeConfig extends ModelConfig {
 
     /**
      * 人脸属性识别模型枚举
@@ -21,16 +22,6 @@ public class FaceAttributeConfig {
      */
     private String modelPath;
 
-    /**
-     * 设备类型
-     */
-    private DeviceEnum device;
-
-    /**
-     * gpu设备ID 当device为GPU时生效
-     */
-    private int gpuId = 0;
-
     /**
      * 是否启用年龄检测
      */

smartjavaai-face/src/main/java/cn/smartjavaai/face/config/FaceDetConfig.java

@@ -1,5 +1,6 @@
 package cn.smartjavaai.face.config;
 
+import cn.smartjavaai.common.config.ModelConfig;
 import cn.smartjavaai.common.enums.DeviceEnum;
 import cn.smartjavaai.face.constant.FaceDetectConstant;
 import cn.smartjavaai.face.enums.FaceDetModelEnum;
@@ -13,7 +14,7 @@ import java.util.Map;
  * @author dwj
  */
 @Data
-public class FaceDetConfig {
+public class FaceDetConfig extends ModelConfig {
 
     /**
      * 人脸检测模型枚举
@@ -36,16 +37,6 @@ public class FaceDetConfig {
      */
     private String modelPath;
 
-    /**
-     * 设备类型
-     */
-    private DeviceEnum device;
-
-    /**
-     * 个性化配置（按模型类型动态解析）
-     */
-    private Map<String, Object> customParams = new HashMap<>();
-
 
     public FaceDetConfig() {
     }
@@ -59,19 +50,4 @@ public class FaceDetConfig {
         this.modelPath = modelPath;
     }
 
-    public <T> T getCustomParam(String key, Class<T> clazz) {
-        Object value = customParams.get(key);
-        if (value == null) return null;
-        return clazz.cast(value);
-    }
-
-    /**
-     * 添加个性化配置项
-     */
-    public void putCustomParam(String key, Object value) {
-        if (customParams == null) {
-            customParams = new HashMap<>();
-        }
-        customParams.put(key, value);
-    }
 }

smartjavaai-face/src/main/java/cn/smartjavaai/face/config/FaceExpressionConfig.java

@@ -1,5 +1,6 @@
 package cn.smartjavaai.face.config;
 
+import cn.smartjavaai.common.config.ModelConfig;
 import cn.smartjavaai.common.enums.DeviceEnum;
 import cn.smartjavaai.face.enums.ExpressionModelEnum;
 import cn.smartjavaai.face.model.facedect.FaceDetModel;
@@ -14,7 +15,7 @@ import java.util.Map;
  * @date 2025/7/1
  */
 @Data
-public class FaceExpressionConfig {
+public class FaceExpressionConfig extends ModelConfig {
 
     /**
      * 模型枚举
@@ -26,11 +27,6 @@ public class FaceExpressionConfig {
      */
     private String modelPath;
 
-    /**
-     * 设备类型
-     */
-    private DeviceEnum device;
-
     /**
      * 人脸检测模型
      */

smartjavaai-face/src/main/java/cn/smartjavaai/face/config/FaceRecConfig.java

@@ -1,5 +1,6 @@
 package cn.smartjavaai.face.config;
 
+import cn.smartjavaai.common.config.ModelConfig;
 import cn.smartjavaai.common.enums.DeviceEnum;
 import cn.smartjavaai.face.constant.FaceDetectConstant;
 import cn.smartjavaai.face.enums.FaceRecModelEnum;
@@ -16,7 +17,7 @@ import java.util.Map;
  * @author dwj
  */
 @Data
-public class FaceRecConfig {
+public class FaceRecConfig extends ModelConfig {
 
     /**
      * 人脸模型枚举
@@ -28,11 +29,6 @@ public class FaceRecConfig {
      */
     private String modelPath;
 
-    /**
-     * 设备类型
-     */
-    private DeviceEnum device;
-
 
     /**
      * 向量数据库配置
@@ -60,10 +56,6 @@ public class FaceRecConfig {
      */
     private FaceDetModel detectModel;
 
-    /**
-     * 个性化配置（按模型类型动态解析）
-     */
-    private Map<String, Object> customParams = new HashMap<>();
 
     public FaceRecConfig() {
     }
@@ -77,19 +69,5 @@ public class FaceRecConfig {
         this.modelPath = modelPath;
     }
 
-    public <T> T getCustomParam(String key, Class<T> clazz) {
-        Object value = customParams.get(key);
-        if (value == null) return null;
-        return clazz.cast(value);
-    }
 
-    /**
-     * 添加个性化配置项
-     */
-    public void putCustomParam(String key, Object value) {
-        if (customParams == null) {
-            customParams = new HashMap<>();
-        }
-        customParams.put(key, value);
-    }
 }

smartjavaai-face/src/main/java/cn/smartjavaai/face/config/LivenessConfig.java

@@ -1,5 +1,6 @@
 package cn.smartjavaai.face.config;
 
+import cn.smartjavaai.common.config.ModelConfig;
 import cn.smartjavaai.common.enums.DeviceEnum;
 import cn.smartjavaai.face.constant.LivenessConstant;
 import cn.smartjavaai.face.enums.LivenessModelEnum;
@@ -15,7 +16,7 @@ import java.util.Map;
  * @author dwj
  */
 @Data
-public class LivenessConfig {
+public class LivenessConfig extends ModelConfig {
 
     /**
      * 活体检测模型枚举
@@ -27,20 +28,12 @@ public class LivenessConfig {
      */
     private String modelPath;
 
-    /**
-     * 设备类型
-     */
-    private DeviceEnum device;
 
     /**
      * 人脸检测模型
      */
     private FaceDetModel detectModel;
 
-    /**
-     * 个性化配置（按模型类型动态解析）
-     */
-    private Map<String, Object> customParams = new HashMap<>();
 
 
     /**
@@ -74,20 +67,4 @@ public class LivenessConfig {
         this.modelPath = modelPath;
     }
 
-    // 可选封装方法，便于类型转换和调用
-    public <T> T getCustomParam(String key, Class<T> clazz) {
-        Object value = customParams.get(key);
-        if (value == null) return null;
-        return clazz.cast(value);
-    }
-
-    /**
-     * 添加个性化配置项
-     */
-    public void putCustomParam(String key, Object value) {
-        if (customParams == null) {
-            customParams = new HashMap<>();
-        }
-        customParams.put(key, value);
-    }
 }

smartjavaai-face/src/main/java/cn/smartjavaai/face/config/QualityConfig.java

@@ -1,5 +1,6 @@
 package cn.smartjavaai.face.config;
 
+import cn.smartjavaai.common.config.ModelConfig;
 import cn.smartjavaai.common.enums.DeviceEnum;
 import cn.smartjavaai.face.constant.LivenessConstant;
 import cn.smartjavaai.face.enums.LivenessModelEnum;
@@ -11,7 +12,7 @@ import lombok.Data;
  * @author dwj
  */
 @Data
-public class QualityConfig {
+public class QualityConfig extends ModelConfig {
 
     /**
      * 活体检测模型枚举
@@ -23,16 +24,6 @@ public class QualityConfig {
      */
     private String modelPath;
 
-    /**
-     * 设备类型
-     */
-    private DeviceEnum device;
-
-    /**
-     * gpu设备ID 当device为GPU时生效
-     */
-    private int gpuId = 0;
-
 
     public QualityConfig() {
     }

smartjavaai-face/src/main/java/cn/smartjavaai/face/model/attribute/Seetaface6FaceAttributeModel.java

@@ -1,5 +1,6 @@
 package cn.smartjavaai.face.model.attribute;
 
+import ai.djl.engine.Engine;
 import cn.smartjavaai.common.entity.*;
 import cn.smartjavaai.common.entity.face.FaceAttribute;
 import cn.smartjavaai.common.entity.face.FaceInfo;
@@ -102,6 +103,19 @@ public class Seetaface6FaceAttributeModel implements FaceAttributeModel {
             this.eyeStateDetectorPool = new EyeStateDetectorPool(eyeStateDetectorPoolConfSetting);
             this.maskDetectorPool = new MaskDetectorPool(maskDetectorPoolConfSetting);
             this.poseEstimatorPool = new PoseEstimatorPool(poseEstimatorPoolConfSetting);
+
+            int predictorPoolSize = config.getPredictorPoolSize();
+            if(config.getPredictorPoolSize() <= 0){
+                predictorPoolSize = Runtime.getRuntime().availableProcessors(); // 默认等于CPU核心数
+            }
+            faceDetectorPool.setMaxTotal(predictorPoolSize);
+            faceLandmarkerPool.setMaxTotal(predictorPoolSize);
+            genderPredictorPool.setMaxTotal(predictorPoolSize);
+            agePredictorPool.setMaxTotal(predictorPoolSize);
+            eyeStateDetectorPool.setMaxTotal(predictorPoolSize);
+            maskDetectorPool.setMaxTotal(predictorPoolSize);
+            poseEstimatorPool.setMaxTotal(predictorPoolSize);
+            log.debug("模型推理器线程池最大数量: " + predictorPoolSize);
         } catch (FileNotFoundException e) {
             throw new FaceException(e);
         }
@@ -471,6 +485,35 @@ public class Seetaface6FaceAttributeModel implements FaceAttributeModel {
         }
     }
 
+
+    public FaceDetectorPool getFaceDetectorPool() {
+        return faceDetectorPool;
+    }
+
+    public GenderPredictorPool getGenderPredictorPool() {
+        return genderPredictorPool;
+    }
+
+    public FaceLandmarkerPool getFaceLandmarkerPool() {
+        return faceLandmarkerPool;
+    }
+
+    public AgePredictorPool getAgePredictorPool() {
+        return agePredictorPool;
+    }
+
+    public EyeStateDetectorPool getEyeStateDetectorPool() {
+        return eyeStateDetectorPool;
+    }
+
+    public MaskDetectorPool getMaskDetectorPool() {
+        return maskDetectorPool;
+    }
+
+    public PoseEstimatorPool getPoseEstimatorPool() {
+        return poseEstimatorPool;
+    }
+
     @Override
     public void close() throws Exception {
         if(Objects.nonNull(faceDetectorPool)){

smartjavaai-face/src/main/java/cn/smartjavaai/face/model/expression/CommonEmotionModel.java

@@ -2,6 +2,7 @@ package cn.smartjavaai.face.model.expression;
 
 import ai.djl.Device;
 import ai.djl.MalformedModelException;
+import ai.djl.engine.Engine;
 import ai.djl.inference.Predictor;
 import ai.djl.modality.Classifications;
 import ai.djl.modality.cv.Image;
@@ -32,6 +33,7 @@ import lombok.extern.slf4j.Slf4j;
 import org.apache.commons.lang3.StringUtils;
 import org.apache.commons.pool2.ObjectPool;
 import org.apache.commons.pool2.impl.GenericObjectPool;
+import org.opencv.core.Mat;
 import org.opencv.face.Face;
 
 import javax.imageio.ImageIO;
@@ -56,7 +58,7 @@ public class CommonEmotionModel implements ExpressionModel{
 
     private ZooModel<Image, Classifications> model;
 
-    private ObjectPool<Predictor<Image, Classifications>> predictorPool;
+    private GenericObjectPool<Predictor<Image, Classifications>> predictorPool;
 
     @Override
     public void loadModel(FaceExpressionConfig config) {
@@ -73,6 +75,14 @@ public class CommonEmotionModel implements ExpressionModel{
         try {
             model = criteria.loadModel();
             this.predictorPool = new GenericObjectPool<>(new PredictorFactory<>(model));
+            int predictorPoolSize = config.getPredictorPoolSize();
+            if(config.getPredictorPoolSize() <= 0){
+                predictorPoolSize = Runtime.getRuntime().availableProcessors(); // 默认等于CPU核心数
+            }
+            predictorPool.setMaxTotal(predictorPoolSize);
+            log.debug("当前设备: " + model.getNDManager().getDevice());
+            log.debug("当前引擎: " + Engine.getInstance().getEngineName());
+            log.debug("模型推理器线程池最大数量: " + predictorPoolSize);
         } catch (IOException | ModelNotFoundException | MalformedModelException e) {
             throw new FaceException("DenseNetEmotionModel模型加载失败", e);
         }
@@ -153,6 +163,7 @@ public class CommonEmotionModel implements ExpressionModel{
             result.setClassifications(classifications);
             faceInfo.setExpressionResult(result);
         }
+        ((Mat)djlImage.getWrappedImage()).release();
         return faceDetectionResponse;
     }
 
@@ -226,6 +237,7 @@ public class CommonEmotionModel implements ExpressionModel{
             result.setClassifications(classifications);
             expressionResults.add(result);
         }
+        ((Mat)djlImage.getWrappedImage()).release();
         return R.ok(expressionResults);
     }
 
@@ -276,6 +288,7 @@ public class CommonEmotionModel implements ExpressionModel{
         FacialExpression expression = FacialExpression.fromLabel(bestClass.getClassName());
         ExpressionResult result = new ExpressionResult(expression, (float)bestClass.getProbability());
         result.setClassifications(classifications);
+        ((Mat)djlImage.getWrappedImage()).release();
         return R.ok(result);
 
     }
@@ -342,7 +355,10 @@ public class CommonEmotionModel implements ExpressionModel{
         return detectTopFace(imageData);
     }
 
-
+    @Override
+    public GenericObjectPool<Predictor<Image, Classifications>> getPool() {
+        return predictorPool;
+    }
 
     @Override
     public void close() {

smartjavaai-face/src/main/java/cn/smartjavaai/face/model/expression/ExpressionModel.java

@@ -1,11 +1,16 @@
 package cn.smartjavaai.face.model.expression;
 
+import ai.djl.inference.Predictor;
+import ai.djl.modality.Classifications;
+import ai.djl.modality.cv.Image;
+import ai.djl.modality.cv.output.DetectedObjects;
 import cn.smartjavaai.common.entity.DetectionRectangle;
 import cn.smartjavaai.common.entity.DetectionResponse;
 import cn.smartjavaai.common.entity.Point;
 import cn.smartjavaai.common.entity.R;
 import cn.smartjavaai.common.entity.face.ExpressionResult;
 import cn.smartjavaai.face.config.FaceExpressionConfig;
+import org.apache.commons.pool2.impl.GenericObjectPool;
 
 import java.awt.image.BufferedImage;
 import java.util.List;
@@ -190,6 +195,9 @@ public interface ExpressionModel extends AutoCloseable{
 
 
 
+    default GenericObjectPool<Predictor<Image, Classifications>> getPool(){
+        throw new UnsupportedOperationException("默认不支持该功能");
+    }
 
 
 }

smartjavaai-face/src/main/java/cn/smartjavaai/face/model/expression/criterial/EmotionCriteriaFactory.java

@@ -26,7 +26,7 @@ public class EmotionCriteriaFactory {
     public static Criteria<Image, Classifications> createCriteria(FaceExpressionConfig config) {
         Device device = null;
         if(!Objects.isNull(config.getDevice())){
-            device = config.getDevice() == DeviceEnum.CPU ? Device.cpu() : Device.gpu();
+            device = config.getDevice() == DeviceEnum.CPU ? Device.cpu() : Device.gpu(config.getGpuId());
         }
         Criteria<Image, Classifications> criteria = null;
         if(config.getModelEnum() == ExpressionModelEnum.DensNet121){
@@ -50,6 +50,7 @@ public class EmotionCriteriaFactory {
                             .optModelPath(Paths.get(config.getModelPath()))
                             .optTranslator(new FrEmotionTranslator(224))
                             .optProgress(new ProgressBar())
+                            .optDevice(device)
                             .build();
         }
         return criteria;

smartjavaai-face/src/main/java/cn/smartjavaai/face/model/facedect/CommonFaceDetModel.java

@@ -24,6 +24,7 @@ import lombok.extern.slf4j.Slf4j;
 import org.apache.commons.lang3.StringUtils;
 import org.apache.commons.pool2.ObjectPool;
 import org.apache.commons.pool2.impl.GenericObjectPool;
+import org.opencv.core.Mat;
 
 import javax.imageio.ImageIO;
 import java.awt.image.BufferedImage;
@@ -43,7 +44,7 @@ import java.util.Objects;
 @Slf4j
 public class CommonFaceDetModel implements FaceDetModel{
 
-    private ObjectPool<Predictor<Image, DetectedObjects>> predictorPool;
+    private GenericObjectPool<Predictor<Image, DetectedObjects>> predictorPool;
 
     private ZooModel<Image, DetectedObjects> model;
 
@@ -58,8 +59,14 @@ public class CommonFaceDetModel implements FaceDetModel{
         try {
             model = criteria.loadModel();
             this.predictorPool = new GenericObjectPool<>(new PredictorFactory<>(model));
+            int predictorPoolSize = config.getPredictorPoolSize();
+            if(config.getPredictorPoolSize() <= 0){
+                predictorPoolSize = Runtime.getRuntime().availableProcessors(); // 默认等于CPU核心数
+            }
+            predictorPool.setMaxTotal(predictorPoolSize);
             log.debug("当前设备: " + model.getNDManager().getDevice());
             log.debug("当前引擎: " + Engine.getInstance().getEngineName());
+            log.debug("模型推理器线程池最大数量: " + predictorPoolSize);
         } catch (IOException | ModelNotFoundException | MalformedModelException e) {
             throw new FaceException("人脸检测模型加载失败", e);
         }
@@ -81,11 +88,16 @@ public class CommonFaceDetModel implements FaceDetModel{
         Image img = null;
         try {
             img = ImageFactory.getInstance().fromFile(Paths.get(imagePath));
+            DetectedObjects detection = detect(img);
+            return R.ok(FaceUtils.convertToDetectionResponse(detection,img));
         } catch (IOException e) {
             throw new FaceException("无效的图片", e);
+        } finally {
+            if (img != null) {
+                ((Mat)img.getWrappedImage()).release();
+            }
         }
-        DetectedObjects detection = detect(img);
-        return R.ok(FaceUtils.convertToDetectionResponse(detection,img));
+
     }
 
     /**
@@ -99,12 +111,17 @@ public class CommonFaceDetModel implements FaceDetModel{
         if(Objects.isNull(imageInputStream)){
             return R.fail(R.Status.INVALID_IMAGE);
         }
+        Image img = null;
         try {
-            Image img = ImageFactory.getInstance().fromInputStream(imageInputStream);
+            img = ImageFactory.getInstance().fromInputStream(imageInputStream);
             DetectedObjects detection = detect(img);
             return R.ok(FaceUtils.convertToDetectionResponse(detection,img));
         } catch (IOException e) {
             throw new FaceException("无效图片输入流", e);
+        } finally {
+            if (img != null) {
+                ((Mat)img.getWrappedImage()).release();
+            }
         }
 
     }
@@ -114,9 +131,19 @@ public class CommonFaceDetModel implements FaceDetModel{
         if(!ImageUtils.isImageValid(image)){
             return R.fail(R.Status.INVALID_IMAGE);
         }
-        Image img = ImageFactory.getInstance().fromImage(OpenCVUtils.image2Mat(image));
-        DetectedObjects detection = detect(img);
-        return R.ok(FaceUtils.convertToDetectionResponse(detection,img));
+        Image img = null;
+        try {
+            img = ImageFactory.getInstance().fromImage(OpenCVUtils.image2Mat(image));
+            DetectedObjects detection = detect(img);
+            return R.ok(FaceUtils.convertToDetectionResponse(detection,img));
+        } catch (Exception e) {
+            throw new FaceException(e);
+        } finally {
+            if (img != null) {
+                ((Mat)img.getWrappedImage()).release();
+            }
+        }
+
     }
 
     @Override
@@ -124,11 +151,7 @@ public class CommonFaceDetModel implements FaceDetModel{
         if(Objects.isNull(imageData)){
             return R.fail(R.Status.INVALID_IMAGE);
         }
-        try {
-            return detect(ImageIO.read(new ByteArrayInputStream(imageData)));
-        } catch (IOException e) {
-            throw new FaceException("错误的图像", e);
-        }
+        return detect(new ByteArrayInputStream(imageData));
     }
 
     @Override
@@ -145,8 +168,9 @@ public class CommonFaceDetModel implements FaceDetModel{
         if(!FileUtils.isFileExists(imagePath)){
             return R.fail(R.Status.FILE_NOT_FOUND);
         }
+        Image img = null;
         try {
-            Image img = ImageFactory.getInstance().fromFile(Paths.get(imagePath));
+            img = ImageFactory.getInstance().fromFile(Paths.get(imagePath));
             DetectedObjects detectedObjects = detect(img);
             if(Objects.isNull(detectedObjects) || detectedObjects.getNumberOfObjects() == 0){
                 return R.fail(R.Status.NO_FACE_DETECTED);
@@ -158,6 +182,10 @@ public class CommonFaceDetModel implements FaceDetModel{
             return R.ok();
         } catch (IOException e) {
             throw new FaceException(e);
+        } finally {
+            if (img != null){
+                ((Mat)img.getWrappedImage()).release();
+            }
         }
     }
 
@@ -181,6 +209,10 @@ public class CommonFaceDetModel implements FaceDetModel{
             return R.ok(ImageIO.read(new ByteArrayInputStream(imageBytes)));
         } catch (IOException e) {
             throw new FaceException("导出图片失败", e);
+        } finally {
+            if (img != null){
+                ((Mat)img.getWrappedImage()).release();
+            }
         }
     }
 
@@ -213,6 +245,11 @@ public class CommonFaceDetModel implements FaceDetModel{
     }
 
 
+    @Override
+    public GenericObjectPool<Predictor<Image, DetectedObjects>> getPool() {
+        return predictorPool;
+    }
+
     @Override
     public void close() {
         try {

smartjavaai-face/src/main/java/cn/smartjavaai/face/model/facedect/FaceDetModel.java

@@ -1,8 +1,13 @@
 package cn.smartjavaai.face.model.facedect;
 
+import ai.djl.inference.Predictor;
+import ai.djl.modality.Classifications;
+import ai.djl.modality.cv.Image;
+import ai.djl.modality.cv.output.DetectedObjects;
 import cn.smartjavaai.common.entity.DetectionResponse;
 import cn.smartjavaai.common.entity.R;
 import cn.smartjavaai.face.config.FaceDetConfig;
+import org.apache.commons.pool2.impl.GenericObjectPool;
 
 import java.awt.image.BufferedImage;
 import java.io.InputStream;
@@ -83,6 +88,10 @@ public interface FaceDetModel extends AutoCloseable{
         throw new UnsupportedOperationException("默认不支持该功能");
     }
 
+    default GenericObjectPool<Predictor<Image, DetectedObjects>> getPool(){
+        throw new UnsupportedOperationException("默认不支持该功能");
+    }
+
 
 
 }

smartjavaai-face/src/main/java/cn/smartjavaai/face/model/facedect/SeetaFace6FaceDetModel.java

@@ -1,5 +1,6 @@
 package cn.smartjavaai.face.model.facedect;
 
+import ai.djl.engine.Engine;
 import cn.smartjavaai.common.entity.DetectionResponse;
 import cn.smartjavaai.common.entity.R;
 import cn.smartjavaai.common.enums.DeviceEnum;
@@ -48,13 +49,9 @@ public class SeetaFace6FaceDetModel implements FaceDetModel{
         String[] faceDetectorModelPath = {config.getModelPath() + File.separator + "face_detector.csta"};
         String[] faceLandmarkerModelPath = {config.getModelPath() + File.separator + "face_landmarker_pts5.csta"};
         SeetaDevice device = SeetaDevice.SEETA_DEVICE_AUTO;
-        int gpuId = 0;
+        int gpuId = config.getGpuId();
         if(Objects.nonNull(config.getDevice())){
             device = config.getDevice() == DeviceEnum.CPU ? SeetaDevice.SEETA_DEVICE_CPU : SeetaDevice.SEETA_DEVICE_GPU;
-            Integer gpuIdValue = config.getCustomParam("gpuId", Integer.class);
-            if(Objects.nonNull(gpuIdValue) && device == SeetaDevice.SEETA_DEVICE_GPU){
-                gpuId = gpuIdValue;
-            }
         }
         try {
             SeetaModelSetting faceDetectorPoolSetting = new SeetaModelSetting(gpuId, faceDetectorModelPath, device);
@@ -65,6 +62,14 @@ public class SeetaFace6FaceDetModel implements FaceDetModel{
 
             this.faceDetectorPool = new FaceDetectorPool(faceDetectorPoolConfSetting);
             this.faceLandmarkerPool = new FaceLandmarkerPool(faceLandmarkerPoolConfSetting);
+
+            int predictorPoolSize = config.getPredictorPoolSize();
+            if(config.getPredictorPoolSize() <= 0){
+                predictorPoolSize = Runtime.getRuntime().availableProcessors(); // 默认等于CPU核心数
+            }
+            faceDetectorPool.setMaxTotal(predictorPoolSize);
+            faceLandmarkerPool.setMaxTotal(predictorPoolSize);
+            log.debug("模型推理器线程池最大数量: " + predictorPoolSize);
         } catch (FileNotFoundException e) {
             throw new FaceException(e);
         }
@@ -212,8 +217,13 @@ public class SeetaFace6FaceDetModel implements FaceDetModel{
     }
 
 
+    public FaceDetectorPool getFaceDetectorPool() {
+        return faceDetectorPool;
+    }
 
-
+    public FaceLandmarkerPool getFaceLandmarkerPool() {
+        return faceLandmarkerPool;
+    }
 
     @Override
     public void close() throws Exception {

smartjavaai-face/src/main/java/cn/smartjavaai/face/model/facedect/criterial/FaceDetCriteriaFactory.java

@@ -31,7 +31,7 @@ public class FaceDetCriteriaFactory {
     public static Criteria<Image, DetectedObjects> createCriteria(FaceDetConfig config) {
         Device device = null;
         if(!Objects.isNull(config.getDevice())){
-            device = config.getDevice() == DeviceEnum.CPU ? Device.cpu() : Device.gpu();
+            device = config.getDevice() == DeviceEnum.CPU ? Device.cpu() : Device.gpu(config.getGpuId());
         }
         Criteria<Image, DetectedObjects> criteria = null;
         if(config.getModelEnum() == FaceDetModelEnum.RETINA_FACE){

smartjavaai-face/src/main/java/cn/smartjavaai/face/model/facerec/CommonFaceRecModel.java

@@ -9,6 +9,7 @@ import ai.djl.ndarray.NDManager;
 import ai.djl.repository.zoo.Criteria;
 import ai.djl.repository.zoo.ModelNotFoundException;
 import ai.djl.repository.zoo.ZooModel;
+import cn.hutool.core.lang.UUID;
 import cn.smartjavaai.common.entity.*;
 import cn.smartjavaai.common.entity.face.FaceInfo;
 import cn.smartjavaai.common.entity.face.FaceSearchResult;
@@ -41,6 +42,7 @@ import lombok.extern.slf4j.Slf4j;
 import org.apache.commons.lang3.StringUtils;
 import org.apache.commons.pool2.ObjectPool;
 import org.apache.commons.pool2.impl.GenericObjectPool;
+import org.opencv.core.Mat;
 
 import javax.imageio.ImageIO;
 import java.awt.image.BufferedImage;
@@ -68,7 +70,7 @@ public class CommonFaceRecModel implements FaceRecModel{
     private static volatile boolean isLoadCompleted = false;
 
 
-    private ObjectPool<Predictor<Image, float[]>> predictorPool;
+    private GenericObjectPool<Predictor<Image, float[]>> predictorPool;
 
 
     private ZooModel<Image, float[]> model;
@@ -101,8 +103,14 @@ public class CommonFaceRecModel implements FaceRecModel{
             model = faceFeatureCriteria.loadModel();
             // 创建池子：每个线程独享 Predictor
             this.predictorPool = new GenericObjectPool<>(new PredictorFactory<>(model));
+            int predictorPoolSize = config.getPredictorPoolSize();
+            if(config.getPredictorPoolSize() <= 0){
+                predictorPoolSize = Runtime.getRuntime().availableProcessors(); // 默认等于CPU核心数
+            }
+            predictorPool.setMaxTotal(predictorPoolSize);
             log.debug("当前设备: " + model.getNDManager().getDevice());
             log.debug("当前引擎: " + Engine.getInstance().getEngineName());
+            log.debug("模型推理器线程池最大数量: " + predictorPoolSize);
         } catch (IOException | ModelNotFoundException | MalformedModelException e) {
             throw new FaceException("模型加载失败", e);
         }
@@ -286,6 +294,7 @@ public class CommonFaceRecModel implements FaceRecModel{
                 faceInfo.setFeature(features);
             }
         }
+        ((Mat)djlImage.getWrappedImage()).release();
         return detectedResult;
     }
 
@@ -350,6 +359,7 @@ public class CommonFaceRecModel implements FaceRecModel{
             }
             features = featureExtraction(subImage);
         }
+        ((Mat)djlImage.getWrappedImage()).release();
         return Objects.isNull(features) ? R.fail(R.Status.Unknown) : R.ok(features);
     }
 
@@ -703,5 +713,8 @@ public class CommonFaceRecModel implements FaceRecModel{
     }
 
 
-
+    @Override
+    public GenericObjectPool<Predictor<Image, float[]>> getPool() {
+        return predictorPool;
+    }
 }

smartjavaai-face/src/main/java/cn/smartjavaai/face/model/facerec/FaceRecModel.java

@@ -1,5 +1,7 @@
 package cn.smartjavaai.face.model.facerec;
 
+import ai.djl.inference.Predictor;
+import ai.djl.modality.cv.Image;
 import cn.smartjavaai.common.entity.DetectionResponse;
 import cn.smartjavaai.common.entity.R;
 import cn.smartjavaai.face.config.FaceRecConfig;
@@ -7,6 +9,7 @@ import cn.smartjavaai.face.entity.FaceRegisterInfo;
 import cn.smartjavaai.face.entity.FaceSearchParams;
 import cn.smartjavaai.common.entity.face.FaceSearchResult;
 import cn.smartjavaai.face.vector.entity.FaceVector;
+import org.apache.commons.pool2.impl.GenericObjectPool;
 
 import java.awt.image.BufferedImage;
 import java.io.InputStream;
@@ -367,4 +370,9 @@ public interface FaceRecModel extends AutoCloseable{
         throw new UnsupportedOperationException("默认不支持该功能");
     }
 
+
+    default GenericObjectPool<Predictor<Image, float[]>> getPool() {
+        throw new UnsupportedOperationException("默认不支持该功能");
+    }
+
 }

smartjavaai-face/src/main/java/cn/smartjavaai/face/model/facerec/SeetaFace6FaceRecModel.java

@@ -1,5 +1,6 @@
 package cn.smartjavaai.face.model.facerec;
 
+import ai.djl.engine.Engine;
 import cn.smartjavaai.common.entity.DetectionInfo;
 import cn.smartjavaai.common.entity.DetectionResponse;
 import cn.smartjavaai.common.entity.R;
@@ -89,13 +90,9 @@ public class SeetaFace6FaceRecModel implements FaceRecModel{
         String[] faceRecognizerModelPath = {config.getModelPath() + File.separator + "face_recognizer.csta"};
         String[] faceLandmarkerModelPath = {config.getModelPath() + File.separator + "face_landmarker_pts5.csta"};
         SeetaDevice device = SeetaDevice.SEETA_DEVICE_AUTO;
-        int gpuId = 0;
+        int gpuId = config.getGpuId();
         if(Objects.nonNull(config.getDevice())){
             device = config.getDevice() == DeviceEnum.CPU ? SeetaDevice.SEETA_DEVICE_CPU : SeetaDevice.SEETA_DEVICE_GPU;
-            Integer gpuIdValue = config.getCustomParam("gpuId", Integer.class);
-            if(Objects.nonNull(gpuIdValue) && device == SeetaDevice.SEETA_DEVICE_GPU){
-                gpuId = gpuIdValue;
-            }
         }
         try {
             SeetaModelSetting faceDetectorPoolSetting = new SeetaModelSetting(gpuId, faceDetectorModelPath, device);
@@ -115,6 +112,16 @@ public class SeetaFace6FaceRecModel implements FaceRecModel{
             this.faceLandmarkerPool = new FaceLandmarkerPool(faceLandmarkerPoolConfSetting);
             this.faceDatabasePool = new FaceDatabasePool(faceDatabasePoolConfSetting);
 
+            int predictorPoolSize = config.getPredictorPoolSize();
+            if(config.getPredictorPoolSize() <= 0){
+                predictorPoolSize = Runtime.getRuntime().availableProcessors(); // 默认等于CPU核心数
+            }
+            faceDetectorPool.setMaxTotal(predictorPoolSize);
+            faceRecognizerPool.setMaxTotal(predictorPoolSize);
+            faceLandmarkerPool.setMaxTotal(predictorPoolSize);
+            faceDatabasePool.setMaxTotal(predictorPoolSize);
+            log.debug("模型推理器线程池最大数量: " + predictorPoolSize);
+
 
             //初始化人脸库
             if(config.getVectorDBConfig() != null && config.getVectorDBConfig().getType() != null){
@@ -739,6 +746,36 @@ public class SeetaFace6FaceRecModel implements FaceRecModel{
         }
     }
 
+    /**
+     * 特征提取
+     * @param image
+     * @return
+     */
+    public float[] featureExtraction(BufferedImage image){
+        FaceRecognizer faceRecognizer = null;
+        try {
+            faceRecognizer = faceRecognizerPool.borrowObject();
+            SeetaImageData imageData = new SeetaImageData(image.getWidth(), image.getHeight(), 3);
+            imageData.data = ImageUtils.getMatrixBGR(image);
+            //提取特征
+            float[] features = new float[faceRecognizer.GetExtractFeatureSize()];
+            faceRecognizer.ExtractCroppedFace(imageData, features);
+            return features;
+        } catch (FaceException e) {
+            throw e;
+        } catch (Exception e) {
+            throw new FaceException("目标检测错误", e);
+        }finally {
+            if (faceRecognizer != null) {
+                try {
+                    faceRecognizerPool.returnObject(faceRecognizer); //归还
+                } catch (Exception e) {
+                    log.warn("归还Predictor失败", e);
+                }
+            }
+        }
+    }
+
     @Override
     public R<float[]> extractTopFaceFeature(String imagePath) {
         if(!FileUtils.isFileExists(imagePath)){
@@ -878,4 +915,19 @@ public class SeetaFace6FaceRecModel implements FaceRecModel{
     }
 
 
+    public FaceDetectorPool getFaceDetectorPool() {
+        return faceDetectorPool;
+    }
+
+    public FaceRecognizerPool getFaceRecognizerPool() {
+        return faceRecognizerPool;
+    }
+
+    public FaceLandmarkerPool getFaceLandmarkerPool() {
+        return faceLandmarkerPool;
+    }
+
+    public FaceDatabasePool getFaceDatabasePool() {
+        return faceDatabasePool;
+    }
 }

smartjavaai-face/src/main/java/cn/smartjavaai/face/model/facerec/criteria/FaceRecCriteriaFactory.java

@@ -27,7 +27,7 @@ public class FaceRecCriteriaFactory {
     public static Criteria<Image, float[]> createCriteria(FaceRecConfig config) {
         Device device = null;
         if(!Objects.isNull(config.getDevice())){
-            device = config.getDevice() == DeviceEnum.CPU ? Device.cpu() : Device.gpu();
+            device = config.getDevice() == DeviceEnum.CPU ? Device.cpu() : Device.gpu(config.getGpuId());
         }
         Criteria<Image, float[]> criteria = null;
         if(config.getModelEnum() == FaceRecModelEnum.FACENET_MODEL){
@@ -58,6 +58,7 @@ public class FaceRecCriteriaFactory {
                     .optTranslator(new FaceFeatureTranslator())
                     .optEngine("PyTorch") // Use PyTorch engine
                     .optProgress(new ProgressBar())
+                    .optDevice(device)
                     .build();
         }else if (config.getModelEnum() == FaceRecModelEnum.INSIGHT_FACE_IRSE50_MODEL){
             if(StringUtils.isBlank(config.getModelPath())){
@@ -73,6 +74,7 @@ public class FaceRecCriteriaFactory {
 //                    .optArgument("resize", "112,112")
                     .optTranslator(new FaceFeatureTranslator())
                     .optEngine("PyTorch") // Use PyTorch engine
+                    .optDevice(device)
                     .optProgress(new ProgressBar())
                     .build();
         }else if (config.getModelEnum() == FaceRecModelEnum.ELASTIC_FACE_MODEL){
@@ -89,6 +91,7 @@ public class FaceRecCriteriaFactory {
 //                    .optArgument("resize", "112,112")
                     .optTranslator(new FaceFeatureTranslator())
                     .optEngine("PyTorch") // Use PyTorch engine
+                    .optDevice(device)
                     .optProgress(new ProgressBar())
                     .build();
         }

smartjavaai-face/src/main/java/cn/smartjavaai/face/model/facerec/translator/FaceFeatureTranslator.java

@@ -31,7 +31,7 @@ public final class FaceFeatureTranslator implements Translator<Image, float[]> {
         NDArray array = input.toNDArray(ctx.getNDManager(), Image.Flag.COLOR);
         Pipeline pipeline = new Pipeline();
         if(input.getWidth() != 112 || input.getHeight() != 112){
-            pipeline.add(new Resize(112));
+            pipeline.add(new Resize(112,112));
         }
         pipeline
                 .add(new ToTensor())

smartjavaai-face/src/main/java/cn/smartjavaai/face/model/liveness/CommonLivenessModel.java

@@ -2,6 +2,7 @@ package cn.smartjavaai.face.model.liveness;
 
 import ai.djl.Device;
 import ai.djl.MalformedModelException;
+import ai.djl.engine.Engine;
 import ai.djl.inference.Predictor;
 import ai.djl.modality.cv.Image;
 import ai.djl.modality.cv.ImageFactory;
@@ -35,6 +36,7 @@ import org.apache.commons.pool2.impl.GenericObjectPool;
 import org.bytedeco.javacv.FFmpegFrameGrabber;
 import org.bytedeco.javacv.Frame;
 import org.bytedeco.javacv.Java2DFrameUtils;
+import org.opencv.core.Mat;
 
 import javax.imageio.ImageIO;
 import java.awt.image.BufferedImage;
@@ -54,7 +56,7 @@ import java.util.*;
 @Slf4j
 public class CommonLivenessModel implements LivenessDetModel{
 
-    protected ObjectPool<Predictor<Image, Float>> predictorPool;
+    protected GenericObjectPool<Predictor<Image, Float>> predictorPool;
 
     protected LivenessConfig config;
 
@@ -78,6 +80,15 @@ public class CommonLivenessModel implements LivenessDetModel{
         } catch (IOException | ModelNotFoundException | MalformedModelException e) {
             throw new FaceException("阿里通义实验室活体检测模型加载失败", e);
         }
+
+        int predictorPoolSize = config.getPredictorPoolSize();
+        if(config.getPredictorPoolSize() <= 0){
+            predictorPoolSize = Runtime.getRuntime().availableProcessors(); // 默认等于CPU核心数
+        }
+        predictorPool.setMaxTotal(predictorPoolSize);
+        log.debug("当前设备: " + model.getNDManager().getDevice());
+        log.debug("当前引擎: " + Engine.getInstance().getEngineName());
+        log.debug("模型推理器线程池最大数量: " + predictorPoolSize);
     }
 
 
@@ -87,6 +98,7 @@ public class CommonLivenessModel implements LivenessDetModel{
             throw new FaceException("图像无效");
         }
         Predictor<Image, Float> predictor = null;
+        Image djlImage = null;
         try {
             predictor = predictorPool.borrowObject();
             //预处理图片
@@ -101,7 +113,7 @@ public class CommonLivenessModel implements LivenessDetModel{
                         .setCenterCropSize(112)
                         .process();
             }
-            Image djlImage = ImageFactory.getInstance().fromImage(OpenCVUtils.image2Mat(processedImage));
+            djlImage = ImageFactory.getInstance().fromImage(OpenCVUtils.image2Mat(processedImage));
             Float result = predictor.predict(djlImage);
             if(result >= config.getRealityThreshold()){
                 return R.ok(new LivenessResult(LivenessStatus.LIVE, result));
@@ -124,6 +136,10 @@ public class CommonLivenessModel implements LivenessDetModel{
                     }
                 }
             }
+            if (djlImage != null){
+                ((Mat)djlImage.getWrappedImage()).release();
+            }
+
         }
     }
 
@@ -403,6 +419,12 @@ public class CommonLivenessModel implements LivenessDetModel{
         return R.fail(R.Status.Unknown);
     }
 
+
+    @Override
+    public GenericObjectPool<Predictor<Image, Float>> getPool() {
+        return predictorPool;
+    }
+
     @Override
     public void close() throws Exception {
         try {

smartjavaai-face/src/main/java/cn/smartjavaai/face/model/liveness/LivenessDetModel.java

@@ -1,11 +1,14 @@
 package cn.smartjavaai.face.model.liveness;
 
+import ai.djl.inference.Predictor;
+import ai.djl.modality.cv.Image;
 import cn.smartjavaai.common.entity.DetectionRectangle;
 import cn.smartjavaai.common.entity.DetectionResponse;
 import cn.smartjavaai.common.entity.Point;
 import cn.smartjavaai.common.entity.R;
 import cn.smartjavaai.common.entity.face.LivenessResult;
 import cn.smartjavaai.face.config.LivenessConfig;
+import org.apache.commons.pool2.impl.GenericObjectPool;
 
 import java.awt.image.BufferedImage;
 import java.io.InputStream;
@@ -292,6 +295,9 @@ public interface LivenessDetModel extends AutoCloseable{
 
 
 
+    default GenericObjectPool<Predictor<Image, Float>> getPool() {
+        throw new UnsupportedOperationException("默认不支持该功能");
+    }
 
 
 

smartjavaai-face/src/main/java/cn/smartjavaai/face/model/liveness/MiniVisionLivenessModel.java

@@ -2,6 +2,7 @@ package cn.smartjavaai.face.model.liveness;
 
 import ai.djl.Device;
 import ai.djl.MalformedModelException;
+import ai.djl.engine.Engine;
 import ai.djl.inference.Predictor;
 import ai.djl.modality.cv.Image;
 import ai.djl.modality.cv.ImageFactory;
@@ -33,6 +34,7 @@ import org.apache.commons.pool2.impl.GenericObjectPool;
 import org.bytedeco.javacv.FFmpegFrameGrabber;
 import org.bytedeco.javacv.Frame;
 import org.bytedeco.javacv.Java2DFrameUtils;
+import org.opencv.core.Mat;
 
 import javax.imageio.ImageIO;
 import java.awt.image.BufferedImage;
@@ -58,9 +60,9 @@ public class MiniVisionLivenessModel extends CommonLivenessModel{
      */
     private static final String SE_MODEL_PATH_KEY = "seModelPath";
 
-    private ObjectPool<Predictor<Image, float[]>> predictorPool;
+    private GenericObjectPool<Predictor<Image, float[]>> predictorPool;
 
-    private ObjectPool<Predictor<Image, float[]>> sePredictorPool;
+    private GenericObjectPool<Predictor<Image, float[]>> sePredictorPool;
 
 
     /**
@@ -87,7 +89,7 @@ public class MiniVisionLivenessModel extends CommonLivenessModel{
         this.config.setRealityThreshold(realityThreshold);
         Device device = null;
         if(!Objects.isNull(config.getDevice())){
-            device = config.getDevice() == DeviceEnum.CPU ? Device.cpu() : Device.gpu();
+            device = config.getDevice() == DeviceEnum.CPU ? Device.cpu() : Device.gpu(config.getGpuId());
         }
         if(StringUtils.isNotBlank(config.getModelPath()) && StringUtils.isBlank(seModelPath)){
             //2.7_80x80_MiniFASNetV2
@@ -109,6 +111,7 @@ public class MiniVisionLivenessModel extends CommonLivenessModel{
                             .optModelPath(Paths.get(config.getModelPath()))
                             .optTranslator(new MiniVisionTranslator())
                             .optProgress(new ProgressBar())
+                            .optDevice(device)
                             .build();
             try {
                 model = criteria.loadModel();
@@ -128,6 +131,7 @@ public class MiniVisionLivenessModel extends CommonLivenessModel{
                             .optModelPath(Paths.get(seModelPath))
                             .optTranslator(new MiniVisionTranslator())
                             .optProgress(new ProgressBar())
+                            .optDevice(device)
                             .build();
             try {
                 seModel = seCriteria.loadModel();
@@ -136,6 +140,16 @@ public class MiniVisionLivenessModel extends CommonLivenessModel{
                 throw new FaceException("MiniFASNetV1SE模型加载失败", e);
             }
         }
+
+        int predictorPoolSize = config.getPredictorPoolSize();
+        if(config.getPredictorPoolSize() <= 0){
+            predictorPoolSize = Runtime.getRuntime().availableProcessors(); // 默认等于CPU核心数
+        }
+        predictorPool.setMaxTotal(predictorPoolSize);
+        sePredictorPool.setMaxTotal(predictorPoolSize);
+        log.debug("当前设备: " + model.getNDManager().getDevice());
+        log.debug("当前引擎: " + Engine.getInstance().getEngineName());
+        log.debug("模型推理器线程池最大数量: " + predictorPoolSize);
     }
 
 
@@ -160,6 +174,8 @@ public class MiniVisionLivenessModel extends CommonLivenessModel{
                 predictor = predictorPool.borrowObject();
                 Image djlImage = ImageFactory.getInstance().fromImage(OpenCVUtils.image2Mat(processedImage));
                 result = predictor.predict(djlImage);
+                ((Mat)djlImage.getWrappedImage()).release();
+
             }
             if(Objects.nonNull(sePredictorPool)){
                 //预处理图片
@@ -172,6 +188,7 @@ public class MiniVisionLivenessModel extends CommonLivenessModel{
                 Image djlImage = ImageFactory.getInstance().fromImage(OpenCVUtils.image2Mat(processedImage));
                 sePredictor = sePredictorPool.borrowObject();
                 seResult = sePredictor.predict(djlImage);
+                ((Mat)djlImage.getWrappedImage()).release();
             }
             if(Objects.isNull(result) && Objects.isNull(seResult)){
                 throw new FaceException("活体检测错误");
@@ -221,6 +238,14 @@ public class MiniVisionLivenessModel extends CommonLivenessModel{
         }
     }
 
+    public GenericObjectPool<Predictor<Image, float[]>> getPredictorPool() {
+        return predictorPool;
+    }
+
+    public GenericObjectPool<Predictor<Image, float[]>> getSePredictorPool() {
+        return sePredictorPool;
+    }
+
     @Override
     public void close() throws Exception {
         try {

smartjavaai-face/src/main/java/cn/smartjavaai/face/model/liveness/Seetaface6LivenessModel.java

@@ -1,5 +1,6 @@
 package cn.smartjavaai.face.model.liveness;
 
+import ai.djl.engine.Engine;
 import cn.smartjavaai.common.entity.*;
 import cn.smartjavaai.common.entity.face.FaceInfo;
 import cn.smartjavaai.common.entity.face.LivenessResult;
@@ -56,13 +57,9 @@ public class Seetaface6LivenessModel implements LivenessDetModel{
         String[] faceAntiSpoofingModelPath = {config.getModelPath() + File.separator + "fas_first.csta",config.getModelPath() + File.separator + "fas_second.csta"};
         String[] faceLandmarkerModelPath = {config.getModelPath() + File.separator + "face_landmarker_pts5.csta"};
         SeetaDevice device = SeetaDevice.SEETA_DEVICE_AUTO;
-        int gpuId = 0;
+        int gpuId = config.getGpuId();
         if(Objects.nonNull(config.getDevice())){
             device = config.getDevice() == DeviceEnum.CPU ? SeetaDevice.SEETA_DEVICE_CPU : SeetaDevice.SEETA_DEVICE_GPU;
-            Integer gpuIdValue = config.getCustomParam("gpuId", Integer.class);
-            if(Objects.nonNull(gpuIdValue) && device == SeetaDevice.SEETA_DEVICE_GPU){
-                gpuId = gpuIdValue;
-            }
         }
 
         try {
@@ -79,6 +76,16 @@ public class Seetaface6LivenessModel implements LivenessDetModel{
             this.faceAntiSpoofingPool = new FaceAntiSpoofingPool(faceAntiSpoofingPoolConfSetting);
             this.faceLandmarkerPool = new FaceLandmarkerPool(faceLandmarkerPoolConfSetting);
 
+            int predictorPoolSize = config.getPredictorPoolSize();
+            if(config.getPredictorPoolSize() <= 0){
+                predictorPoolSize = Runtime.getRuntime().availableProcessors(); // 默认等于CPU核心数
+            }
+
+            faceDetectorPool.setMaxTotal(predictorPoolSize);
+            faceAntiSpoofingPool.setMaxTotal(predictorPoolSize);
+            faceLandmarkerPool.setMaxTotal(predictorPoolSize);
+            log.debug("模型推理器线程池最大数量: " + predictorPoolSize);
+
             //初始化模型参数
             initConfig();
         } catch (FileNotFoundException e) {
@@ -550,6 +557,18 @@ public class Seetaface6LivenessModel implements LivenessDetModel{
         return R.fail(1000, "有效帧数量不足，无法完成活体检测");
     }
 
+    public FaceDetectorPool getFaceDetectorPool() {
+        return faceDetectorPool;
+    }
+
+    public FaceAntiSpoofingPool getFaceAntiSpoofingPool() {
+        return faceAntiSpoofingPool;
+    }
+
+    public FaceLandmarkerPool getFaceLandmarkerPool() {
+        return faceLandmarkerPool;
+    }
+
     @Override
     public void close() throws Exception {
         try {

smartjavaai-face/src/main/java/cn/smartjavaai/face/model/liveness/criterial/LivenessCriteriaFactory.java

@@ -31,7 +31,7 @@ public class LivenessCriteriaFactory {
     public static Criteria<Image, Float> createCriteria(LivenessConfig config) {
         Device device = null;
         if(!Objects.isNull(config.getDevice())){
-            device = config.getDevice() == DeviceEnum.CPU ? Device.cpu() : Device.gpu();
+            device = config.getDevice() == DeviceEnum.CPU ? Device.cpu() : Device.gpu(config.getGpuId());
         }
         Criteria<Image, Float> criteria = null;
         if(config.getModelEnum() == LivenessModelEnum.IIC_FL_MODEL){

smartjavaai-face/src/main/java/cn/smartjavaai/face/model/quality/Seetaface6QualityModel.java

@@ -1,5 +1,6 @@
 package cn.smartjavaai.face.model.quality;
 
+import ai.djl.engine.Engine;
 import cn.smartjavaai.common.entity.*;
 import cn.smartjavaai.common.enums.DeviceEnum;
 import cn.smartjavaai.common.utils.FileUtils;
@@ -74,6 +75,8 @@ public class Seetaface6QualityModel implements FaceQualityModel {
      */
     private QualityOfResolutionPool qualityOfResolutionPool;
 
+    int predictorPoolSize = 0;
+
 
     @Override
     public void loadModel(QualityConfig config) {
@@ -84,6 +87,11 @@ public class Seetaface6QualityModel implements FaceQualityModel {
         //加载依赖库
         NativeLoader.loadNativeLibraries(device);
         this.config = config;
+        int predictorPoolSize = config.getPredictorPoolSize();
+        if(config.getPredictorPoolSize() <= 0){
+            predictorPoolSize = Runtime.getRuntime().availableProcessors(); // 默认等于CPU核心数
+        }
+        log.debug("模型推理器线程池最大数量: " + predictorPoolSize);
         log.debug("Loading seetaFace6 library successfully.");
     }
 
@@ -104,6 +112,7 @@ public class Seetaface6QualityModel implements FaceQualityModel {
         try {
             if(Objects.isNull(this.qualityOfBrightnessPool)){
                 this.qualityOfBrightnessPool = new QualityOfBrightnessPool(new SeetaConfSetting());
+                qualityOfBrightnessPool.setMaxTotal(predictorPoolSize);
             }
             qualityOfBrightness = qualityOfBrightnessPool.borrowObject();
             SeetaImageData imageData = new SeetaImageData(image.getWidth(), image.getHeight(), 3);
@@ -170,6 +179,7 @@ public class Seetaface6QualityModel implements FaceQualityModel {
         try {
             if(Objects.isNull(this.qualityOfClarityPool)){
                 this.qualityOfClarityPool = new QualityOfClarityPool(new SeetaConfSetting());
+                qualityOfClarityPool.setMaxTotal(predictorPoolSize);
             }
             qualityOfClarity = qualityOfClarityPool.borrowObject();
             SeetaImageData imageData = new SeetaImageData(image.getWidth(), image.getHeight(), 3);
@@ -236,6 +246,7 @@ public class Seetaface6QualityModel implements FaceQualityModel {
         try {
             if(Objects.isNull(this.qualityOfIntegrityPool)){
                 this.qualityOfIntegrityPool = new QualityOfIntegrityPool(new SeetaConfSetting());
+                qualityOfIntegrityPool.setMaxTotal(predictorPoolSize);
             }
             qualityOfIntegrity = qualityOfIntegrityPool.borrowObject();
             SeetaImageData imageData = new SeetaImageData(image.getWidth(), image.getHeight(), 3);
@@ -302,6 +313,7 @@ public class Seetaface6QualityModel implements FaceQualityModel {
         try {
             if(Objects.isNull(this.qualityOfPosePool)){
                 this.qualityOfPosePool = new QualityOfPosePool(new SeetaConfSetting());
+                qualityOfPosePool.setMaxTotal(predictorPoolSize);
             }
             qualityOfPose = qualityOfPosePool.borrowObject();
             SeetaImageData imageData = new SeetaImageData(image.getWidth(), image.getHeight(), 3);
@@ -368,6 +380,7 @@ public class Seetaface6QualityModel implements FaceQualityModel {
         try {
             if(Objects.isNull(this.qualityOfResolutionPool)){
                 this.qualityOfResolutionPool = new QualityOfResolutionPool(new SeetaConfSetting());
+                qualityOfResolutionPool.setMaxTotal(predictorPoolSize);
             }
             qualityOfResolution = qualityOfResolutionPool.borrowObject();
             SeetaImageData imageData = new SeetaImageData(image.getWidth(), image.getHeight(), 3);
@@ -438,6 +451,7 @@ public class Seetaface6QualityModel implements FaceQualityModel {
                 }
                 SeetaConfSetting setting = getClarityMLSetting();
                 this.qualityOfLBNPool = new QualityOfLBNPool(setting);
+                qualityOfLBNPool.setMaxTotal(predictorPoolSize);
             }
             qualityOfLBN = qualityOfLBNPool.borrowObject();
             SeetaImageData imageData = new SeetaImageData(image.getWidth(), image.getHeight(), 3);
@@ -509,6 +523,7 @@ public class Seetaface6QualityModel implements FaceQualityModel {
                 }
                 SeetaConfSetting setting = getPoseMLSetting();
                 this.qualityOfPoseExPool = new QualityOfPoseExPool(setting);
+                qualityOfPoseExPool.setMaxTotal(predictorPoolSize);
             }
             qualityOfPoseEx = qualityOfPoseExPool.borrowObject();
             SeetaImageData imageData = new SeetaImageData(image.getWidth(), image.getHeight(), 3);
@@ -628,18 +643,23 @@ public class Seetaface6QualityModel implements FaceQualityModel {
         try {
             if(Objects.isNull(this.qualityOfBrightnessPool)){
                 this.qualityOfBrightnessPool = new QualityOfBrightnessPool(new SeetaConfSetting());
+                qualityOfBrightnessPool.setMaxTotal(predictorPoolSize);
             }
             if(Objects.isNull(this.qualityOfClarityPool)){
                 this.qualityOfClarityPool = new QualityOfClarityPool(new SeetaConfSetting());
+                qualityOfClarityPool.setMaxTotal(predictorPoolSize);
             }
             if(Objects.isNull(this.qualityOfIntegrityPool)){
                 this.qualityOfIntegrityPool = new QualityOfIntegrityPool(new SeetaConfSetting());
+                qualityOfIntegrityPool.setMaxTotal(predictorPoolSize);
             }
             if(Objects.isNull(this.qualityOfPosePool)){
                 this.qualityOfPosePool = new QualityOfPosePool(new SeetaConfSetting());
+                qualityOfPosePool.setMaxTotal(predictorPoolSize);
             }
             if(Objects.isNull(this.qualityOfResolutionPool)){
                 this.qualityOfResolutionPool = new QualityOfResolutionPool(new SeetaConfSetting());
+                qualityOfResolutionPool.setMaxTotal(predictorPoolSize);
             }
             FaceQualitySummary summary = new FaceQualitySummary();
             qualityOfBrightness = qualityOfBrightnessPool.borrowObject();
@@ -690,6 +710,34 @@ public class Seetaface6QualityModel implements FaceQualityModel {
         }
     }
 
+    public QualityOfBrightnessPool getQualityOfBrightnessPool() {
+        return qualityOfBrightnessPool;
+    }
+
+    public QualityOfClarityPool getQualityOfClarityPool() {
+        return qualityOfClarityPool;
+    }
+
+    public QualityOfLBNPool getQualityOfLBNPool() {
+        return qualityOfLBNPool;
+    }
+
+    public QualityOfIntegrityPool getQualityOfIntegrityPool() {
+        return qualityOfIntegrityPool;
+    }
+
+    public QualityOfPosePool getQualityOfPosePool() {
+        return qualityOfPosePool;
+    }
+
+    public QualityOfPoseExPool getQualityOfPoseExPool() {
+        return qualityOfPoseExPool;
+    }
+
+    public QualityOfResolutionPool getQualityOfResolutionPool() {
+        return qualityOfResolutionPool;
+    }
+
     @Override
     public void close() throws Exception {
         if(Objects.nonNull(qualityOfBrightnessPool)){

smartjavaai-face/src/main/java/cn/smartjavaai/face/preprocess/DJLImagePreprocessor.java

@@ -60,15 +60,13 @@ public class DJLImagePreprocessor {
     // 处理流程
     public Image process() {
         Image result = image;
-
-        if (enableCrop) {
-            result = result.getSubImage(cropRect.x, cropRect.y, cropRect.width, cropRect.height);
-        }
-
-        if (enableAffine) {
+        if(enableAffine){
             result = warpAffine(keyPoints, affineTargetWidth, affineTargetHeight);
+        }else {
+            if(enableCrop){
+                result = result.getSubImage(cropRect.x, cropRect.y, cropRect.width, cropRect.height);
+            }
         }
-
         return result;
     }
 

smartjavaai-face/src/main/java/cn/smartjavaai/face/vector/config/MilvusConfig.java

@@ -62,6 +62,16 @@ public class MilvusConfig extends VectorDBConfig {
      */
     private boolean useMemoryCache = true;
 
+    /**
+     * 用户名
+     */
+    private String username;
+
+    /**
+     * 密码
+     */
+    private String password;
+
     /**
      * 构造函数
      */

smartjavaai-face/src/main/java/cn/smartjavaai/face/vector/core/MilvusClient.java

@@ -49,10 +49,13 @@ public class MilvusClient implements VectorDBClient {
     @Override
     public void initialize() {
         try {
-            ConnectParam connectParam = ConnectParam.newBuilder()
+            ConnectParam.Builder builder = ConnectParam.newBuilder()
                     .withHost(config.getHost())
-                    .withPort(config.getPort())
-                    .build();
+                    .withPort(config.getPort());
+            if (StringUtils.isNotBlank(config.getUsername()) && StringUtils.isNotBlank(config.getPassword())) {
+                builder.withAuthorization(config.getUsername(), config.getPassword());
+            }
+            ConnectParam connectParam = builder.build();
             serviceClient = new MilvusServiceClient(connectParam);
             collectionName = StringUtils.isNotBlank(config.getCollectionName()) ? config.getCollectionName() : VectorDBConstants.Defaults.DEFAULT_COLLECTION_NAME;
             createCollection(collectionName, config.getDimension());

smartjavaai-objectdetection/pom.xml

@@ -6,11 +6,11 @@
     <parent>
         <groupId>cn.smartjavaai</groupId>
         <artifactId>smartjavaai-parent</artifactId>
-        <version>1.0.20</version>
+        <version>1.0.22</version>
     </parent>
 
     <artifactId>smartjavaai-objectdetection</artifactId>
-    <version>1.0.20</version>
+    <version>1.0.22</version>
     <name>smartjavaai-objectdetection</name>
     <description>SmartJavaAI</description>
     <url>https://github.com/geekwenjie/SmartJavaAI</url>

smartjavaai-objectdetection/src/main/java/cn/smartjavaai/objectdetection/config/DetectorModelConfig.java

@@ -1,11 +1,13 @@
 package cn.smartjavaai.objectdetection.config;
 
+import cn.smartjavaai.common.config.ModelConfig;
 import cn.smartjavaai.common.enums.DeviceEnum;
 import cn.smartjavaai.objectdetection.constant.DetectorConstant;
 import cn.smartjavaai.objectdetection.enums.DetectorModelEnum;
 import lombok.Data;
 
 import java.util.HashMap;
+import java.util.List;
 import java.util.Map;
 
 /**
@@ -15,7 +17,7 @@ import java.util.Map;
  * @date 2025/4/4
  */
 @Data
-public class DetectorModelConfig {
+public class DetectorModelConfig extends ModelConfig {
 
     /**
      * 模型
@@ -27,10 +29,6 @@ public class DetectorModelConfig {
      */
     private float threshold = DetectorConstant.DEFAULT_THRESHOLD;
 
-    /**
-     * 设备类型
-     */
-    private DeviceEnum device;
 
     /**
      * 模型路径
@@ -45,35 +43,27 @@ public class DetectorModelConfig {
     private int maxBox;
 
     /**
-     * 个性化配置
+     * 允许的类列表
      */
-    private Map<String, Object> customParams = new HashMap<>();
+    private List<String> allowedClasses;
+
+    /**
+     * 检测结果数量
+     */
+    private int topK;
+
 
     public DetectorModelConfig() {
     }
 
     public DetectorModelConfig(DetectorModelEnum modelEnum, DeviceEnum device) {
         this.modelEnum = modelEnum;
-        this.device = device;
+        setDevice(device);
     }
 
     public DetectorModelConfig(DetectorModelEnum modelEnum) {
         this.modelEnum = modelEnum;
     }
 
-    public <T> T getCustomParam(String key, Class<T> clazz) {
-        Object value = customParams.get(key);
-        if (value == null) return null;
-        return clazz.cast(value);
-    }
 
-    /**
-     * 添加个性化配置项
-     */
-    public void putCustomParam(String key, Object value) {
-        if (customParams == null) {
-            customParams = new HashMap<>();
-        }
-        customParams.put(key, value);
-    }
 }

smartjavaai-objectdetection/src/main/java/cn/smartjavaai/objectdetection/criteria/DJLModelCriteriaBuilder.java

@@ -25,7 +25,7 @@ public class DJLModelCriteriaBuilder implements CriteriaBuilderStrategy {
     public Criteria<Image, DetectedObjects> buildCriteria(DetectorModelConfig config) {
         Device device = null;
         if(!Objects.isNull(config.getDevice())){
-            device = config.getDevice() == DeviceEnum.CPU ? Device.cpu() : Device.gpu();
+            device = config.getDevice() == DeviceEnum.CPU ? Device.cpu() : Device.gpu(config.getGpuId());
         }
         Criteria<Image, DetectedObjects> criteria = Criteria.builder()
                 .optApplication(Application.CV.OBJECT_DETECTION)

smartjavaai-objectdetection/src/main/java/cn/smartjavaai/objectdetection/criteria/YoloCriteriaBuilder.java

@@ -25,7 +25,7 @@ public class YoloCriteriaBuilder implements CriteriaBuilderStrategy {
     public Criteria<Image, DetectedObjects> buildCriteria(DetectorModelConfig config) {
         Device device = null;
         if(!Objects.isNull(config.getDevice())){
-            device = config.getDevice() == DeviceEnum.CPU ? Device.cpu() : Device.gpu();
+            device = config.getDevice() == DeviceEnum.CPU ? Device.cpu() : Device.gpu(config.getGpuId());
         }
 
         Map<String, Object> customParams = getDefaultConfig();

smartjavaai-objectdetection/src/main/java/cn/smartjavaai/objectdetection/model/DetectorModel.java

@@ -1,16 +1,22 @@
 package cn.smartjavaai.objectdetection.model;
 
 import ai.djl.MalformedModelException;
+import ai.djl.engine.Engine;
 import ai.djl.inference.Predictor;
 import ai.djl.modality.cv.Image;
 import ai.djl.modality.cv.ImageFactory;
+import ai.djl.modality.cv.output.BoundingBox;
 import ai.djl.modality.cv.output.DetectedObjects;
 import ai.djl.modality.cv.translator.YoloV8TranslatorFactory;
 import ai.djl.repository.zoo.Criteria;
 import ai.djl.repository.zoo.ModelNotFoundException;
 import ai.djl.repository.zoo.ZooModel;
 import ai.djl.training.util.ProgressBar;
+import cn.smartjavaai.common.entity.DetectionInfo;
+import cn.smartjavaai.common.entity.DetectionRectangle;
 import cn.smartjavaai.common.entity.DetectionResponse;
+import cn.smartjavaai.common.entity.Point;
+import cn.smartjavaai.common.entity.face.FaceInfo;
 import cn.smartjavaai.common.pool.PredictorFactory;
 import cn.smartjavaai.common.utils.FileUtils;
 import cn.smartjavaai.common.utils.ImageUtils;
@@ -21,39 +27,53 @@ import cn.smartjavaai.objectdetection.criteria.CriteriaBuilderFactory;
 import cn.smartjavaai.objectdetection.exception.DetectionException;
 import cn.smartjavaai.objectdetection.utils.DetectorUtils;
 import lombok.extern.slf4j.Slf4j;
+import org.apache.commons.collections.CollectionUtils;
 import org.apache.commons.lang3.StringUtils;
 import org.apache.commons.pool2.ObjectPool;
 import org.apache.commons.pool2.impl.GenericObjectPool;
+import org.opencv.core.Mat;
 
 import javax.imageio.ImageIO;
 import java.awt.image.BufferedImage;
 import java.io.*;
 import java.nio.file.Paths;
+import java.util.ArrayList;
+import java.util.Iterator;
+import java.util.List;
 import java.util.Objects;
+import java.util.stream.Collectors;
 
 /**
  * 目标检测模型
  * @author dwj
- * @date 2025/4/4
  */
 @Slf4j
 public class DetectorModel implements AutoCloseable{
 
     private ZooModel<Image, DetectedObjects> model;
 
-    private ObjectPool<Predictor<Image, DetectedObjects>> predictorPool;
+    private GenericObjectPool<Predictor<Image, DetectedObjects>> predictorPool;
+
+    private DetectorModelConfig config;
 
     public void loadModel(DetectorModelConfig config){
         if(Objects.isNull(config.getModelEnum())){
             throw new DetectionException("未配置模型枚举");
         }
         Criteria<Image, DetectedObjects> criteria = CriteriaBuilderFactory.createCriteria(config);
-
+        this.config = config;
         try {
             model = criteria.loadModel();
             // 创建池子：每个线程独享 Predictor
             this.predictorPool = new GenericObjectPool<>(new PredictorFactory<>(model));
+            int predictorPoolSize = config.getPredictorPoolSize();
+            if(config.getPredictorPoolSize() <= 0){
+                predictorPoolSize = Runtime.getRuntime().availableProcessors(); // 默认等于CPU核心数
+            }
+            predictorPool.setMaxTotal(predictorPoolSize);
             log.debug("当前设备: " + model.getNDManager().getDevice());
+            log.debug("当前引擎: " + Engine.getInstance().getEngineName());
+            log.debug("模型推理器线程池最大数量: " + predictorPoolSize);
         } catch (IOException | ModelNotFoundException | MalformedModelException e) {
             throw new DetectionException("模型加载失败", e);
         }
@@ -72,11 +92,16 @@ public class DetectorModel implements AutoCloseable{
         Image image = null;
         try {
             image = ImageFactory.getInstance().fromFile(Paths.get(imagePath));
-        } catch (IOException e) {
-            throw new DetectionException("图片转换错误", e);
+            DetectedObjects detectedObjects = detect(image);
+            return DetectorUtils.convertToDetectionResponse(detectedObjects, image);
+        } catch (Exception e) {
+            throw new DetectionException(e);
+        } finally {
+            if (image != null){
+                ((Mat)image.getWrappedImage()).release();
+            }
         }
-        DetectedObjects detectedObjects = detect(image);
-        return DetectorUtils.convertToDetectionResponse(detectedObjects, image);
+
     }
 
 
@@ -89,8 +114,9 @@ public class DetectorModel implements AutoCloseable{
         if(!FileUtils.isFileExists(imagePath)){
             throw new DetectionException("图像文件不存在");
         }
+        Image img = null;
         try {
-            Image img = ImageFactory.getInstance().fromFile(Paths.get(imagePath));
+            img = ImageFactory.getInstance().fromFile(Paths.get(imagePath));
             DetectedObjects detectedObjects = detect(img);
             img.drawBoundingBoxes(detectedObjects);
             ByteArrayOutputStream outputStream = new ByteArrayOutputStream();
@@ -98,6 +124,10 @@ public class DetectorModel implements AutoCloseable{
             img.save(new FileOutputStream(Paths.get(outputPath).toAbsolutePath().toString()), "png");
         } catch (IOException e) {
             throw new DetectionException(e);
+        } finally {
+            if (img != null){
+                ((Mat)img.getWrappedImage()).release();
+            }
         }
     }
 
@@ -129,9 +159,19 @@ public class DetectorModel implements AutoCloseable{
         if(!ImageUtils.isImageValid(image)){
             throw new DetectionException("图像无效");
         }
-        Image img = ImageFactory.getInstance().fromImage(OpenCVUtils.image2Mat(image));
-        DetectedObjects detectedObjects = detect(img);
-        return DetectorUtils.convertToDetectionResponse(detectedObjects, img);
+        Image img = null;
+        try {
+            img = ImageFactory.getInstance().fromImage(OpenCVUtils.image2Mat(image));
+            DetectedObjects detectedObjects = detect(img);
+            return DetectorUtils.convertToDetectionResponse(detectedObjects, img);
+        } catch (Exception e) {
+            throw new DetectionException(e);
+        } finally {
+            if (img != null) {
+                ((Mat)img.getWrappedImage()).release();
+            }
+        }
+
     }
 
     /**
@@ -155,6 +195,10 @@ public class DetectorModel implements AutoCloseable{
             return ImageIO.read(new ByteArrayInputStream(imageBytes));
         } catch (IOException e) {
             throw new DetectionException("导出图片失败", e);
+        } finally {
+            if (img != null) {
+                ((Mat)img.getWrappedImage()).release();
+            }
         }
     }
 
@@ -163,11 +207,13 @@ public class DetectorModel implements AutoCloseable{
      * @param image
      * @return
      */
-    private DetectedObjects detect(Image image){
+    public DetectedObjects detect(Image image){
         Predictor<Image, DetectedObjects> predictor = null;
         try {
             predictor = predictorPool.borrowObject();
-            return predictor.predict(image);
+            DetectedObjects detectedObjects = predictor.predict(image);
+            detectedObjects = filterDetections(detectedObjects);
+            return detectedObjects;
         } catch (Exception e) {
             throw new DetectionException("目标检测错误", e);
         }finally {
@@ -187,6 +233,50 @@ public class DetectorModel implements AutoCloseable{
         }
     }
 
+    /**
+     * 筛选检测结果
+     * @param detectedObjects
+     * @return
+     */
+    private DetectedObjects filterDetections(DetectedObjects detectedObjects) {
+        if (Objects.isNull(detectedObjects) || detectedObjects.getNumberOfObjects() == 0) {
+            return detectedObjects;
+        }
+        List<DetectedObjects.DetectedObject> items = detectedObjects.items();
+        // 按照允许的类别进行过滤
+        List<DetectedObjects.DetectedObject> filtered = new ArrayList<>();
+        //过滤类别
+        if(!CollectionUtils.isEmpty(config.getAllowedClasses())){
+            for (DetectedObjects.DetectedObject obj : items) {
+                if(config.getAllowedClasses().contains(obj.getClassName())){
+                    filtered.add(obj);
+                }
+            }
+        }else{
+            filtered = items;
+        }
+        // 按照概率进行排序
+        filtered.sort((o1, o2) -> Double.compare(o2.getProbability(), o1.getProbability()));
+        if(config.getTopK() > 0 && filtered.size() > config.getTopK()){
+            filtered = filtered.subList(0, config.getTopK());
+        }
+        // 构建新的 DetectedObjects 返回
+        List<String> names = new ArrayList<>();
+        List<Double> probs = new ArrayList<>();
+        List<BoundingBox> boxes = new ArrayList<>();
+
+        for (DetectedObjects.DetectedObject obj : filtered) {
+            names.add(obj.getClassName());
+            probs.add(obj.getProbability());
+            boxes.add(obj.getBoundingBox());
+        }
+        return new DetectedObjects(names, probs, boxes);
+    }
+
+    public GenericObjectPool<Predictor<Image, DetectedObjects>> getPool() {
+        return predictorPool;
+    }
+
 
     /**
      * 显式释放资源

smartjavaai-ocr/pom.xml

@@ -6,7 +6,7 @@
     <parent>
         <groupId>cn.smartjavaai</groupId>
         <artifactId>smartjavaai-parent</artifactId>
-        <version>1.0.20</version>
+        <version>1.0.22</version>
     </parent>
 
     <artifactId>smartjavaai-ocr</artifactId>
@@ -42,7 +42,7 @@
         </dependency>
     </dependencies>
 
-    <version>1.0.20</version>
+    <version>1.0.22</version>
     <name>smartjavaai-ocr</name>
     <description>SmartJavaAI</description>
     <url>https://github.com/geekwenjie/SmartJavaAI</url>

smartjavaai-ocr/src/main/java/cn/smartjavaai/ocr/config/PlateDetModelConfig.java

@@ -0,0 +1,41 @@
+package cn.smartjavaai.ocr.config;
+
+import cn.smartjavaai.common.config.ModelConfig;
+import cn.smartjavaai.ocr.enums.CommonDetModelEnum;
+import cn.smartjavaai.ocr.enums.PlateDetModelEnum;
+import lombok.Data;
+
+/**
+ * 车牌检测模型配置
+ * @author dwj
+ */
+@Data
+public class PlateDetModelConfig extends ModelConfig {
+
+    /**
+     * 模型
+     */
+    private PlateDetModelEnum modelEnum;
+
+    /**
+     * 检测模型路径
+     */
+    private String modelPath;
+
+    /**
+     * 置信度阈值
+     */
+    private float confidenceThreshold;
+
+    /**
+     * iou阈值
+     */
+    private float iouThreshold;
+
+    /**
+     * 检测结果数量
+     */
+    private int topK;
+
+
+}

smartjavaai-ocr/src/main/java/cn/smartjavaai/ocr/config/PlateRecModelConfig.java

@@ -0,0 +1,47 @@
+package cn.smartjavaai.ocr.config;
+
+import cn.smartjavaai.common.config.ModelConfig;
+import cn.smartjavaai.ocr.enums.PlateDetModelEnum;
+import cn.smartjavaai.ocr.enums.PlateRecModelEnum;
+import cn.smartjavaai.ocr.model.plate.PlateDetModel;
+import lombok.Data;
+
+/**
+ * 车牌识别模型配置
+ * @author dwj
+ */
+@Data
+public class PlateRecModelConfig extends ModelConfig {
+
+    /**
+     * 模型
+     */
+    private PlateRecModelEnum modelEnum;
+
+    /**
+     * 检测模型路径
+     */
+    private String modelPath;
+
+    /**
+     * 车牌检测模型
+     */
+    private PlateDetModel plateDetModel;
+
+    /**
+     * 置信度阈值
+     */
+    private float confidenceThreshold;
+
+    /**
+     * iou阈值
+     */
+    private float iouThreshold;
+
+    /**
+     * 检测结果数量
+     */
+    private int topK;
+
+
+}

smartjavaai-ocr/src/main/java/cn/smartjavaai/ocr/entity/PlateInfo.java

@@ -0,0 +1,45 @@
+package cn.smartjavaai.ocr.entity;
+
+import cn.smartjavaai.common.entity.DetectionRectangle;
+import cn.smartjavaai.ocr.enums.PlateType;
+import lombok.Data;
+
+/**
+ * 车牌识别信息
+ * @author dwj
+ */
+@Data
+public class PlateInfo {
+
+    /**
+     * 车牌类型
+     */
+    private PlateType plateType;
+
+    /**
+     * 车牌号码
+     */
+    private String plateNumber;
+
+    /**
+     * 车牌颜色
+     */
+    private String plateColor;
+
+    /**
+     * 检测位置信息
+     */
+    private DetectionRectangle detectionRectangle;
+
+    /**
+     * 车牌4角坐标
+     */
+    private OcrBox box;
+
+    /**
+     * 检测得分
+     */
+    private float score;
+
+
+}

smartjavaai-ocr/src/main/java/cn/smartjavaai/ocr/entity/PlateResult.java

@@ -0,0 +1,35 @@
+package cn.smartjavaai.ocr.entity;
+
+import lombok.Data;
+
+/**
+ * @author dwj
+ */
+@Data
+public class PlateResult {
+
+    /**
+     * 车牌号码
+     */
+    private String plateNo;
+
+    /**
+     * 车牌颜色
+     */
+    private String plateColor;
+
+    public PlateResult(String plateNo, String plateColor) {
+        this.plateNo = plateNo;
+        this.plateColor = plateColor;
+    }
+
+
+    @Override
+    public String toString() {
+        return "PlateResult{" +
+                "plateNo='" + plateNo + '\'' +
+                ", plateColor='" + plateColor + '\'' +
+                '}';
+    }
+}
+

smartjavaai-ocr/src/main/java/cn/smartjavaai/ocr/enums/PlateDetModelEnum.java

@@ -0,0 +1,28 @@
+package cn.smartjavaai.ocr.enums;
+
+/**
+ * 车牌检测模型枚举
+ * @author dwj
+ */
+public enum PlateDetModelEnum {
+
+    YOLOV5,
+
+    YOLOV7;
+
+
+    /**
+     * 根据名称获取枚举 (忽略大小写和下划线变体)
+     */
+    public static PlateDetModelEnum fromName(String name) {
+        String formatted = name.trim().toUpperCase().replaceAll("[-_]", "");
+        for (PlateDetModelEnum model : values()) {
+            if (model.name().replaceAll("_", "").equals(formatted)) {
+                return model;
+            }
+        }
+        throw new IllegalArgumentException("未知模型名称: " + name);
+    }
+
+
+}

smartjavaai-ocr/src/main/java/cn/smartjavaai/ocr/enums/PlateRecModelEnum.java

@@ -0,0 +1,26 @@
+package cn.smartjavaai.ocr.enums;
+
+/**
+ * 车牌识别模型枚举
+ * @author dwj
+ */
+public enum PlateRecModelEnum {
+
+    PLATE_REC_CRNN;
+
+
+    /**
+     * 根据名称获取枚举 (忽略大小写和下划线变体)
+     */
+    public static PlateRecModelEnum fromName(String name) {
+        String formatted = name.trim().toUpperCase().replaceAll("[-_]", "");
+        for (PlateRecModelEnum model : values()) {
+            if (model.name().replaceAll("_", "").equals(formatted)) {
+                return model;
+            }
+        }
+        throw new IllegalArgumentException("未知模型名称: " + name);
+    }
+
+
+}

smartjavaai-ocr/src/main/java/cn/smartjavaai/ocr/enums/PlateType.java

@@ -0,0 +1,45 @@
+package cn.smartjavaai.ocr.enums;
+
+/**
+ * @author dwj
+ */
+public enum PlateType {
+
+    SINGLE("single", "单层"),
+    DOUBLE("double", "双层"),
+    UNKNOWN("unknown", "未知");
+
+    private final String className;
+    private final String description;
+
+    PlateType(String className, String description) {
+        this.className = className;
+        this.description = description;
+    }
+
+    public String getClassName() {
+        return className;
+    }
+
+    public String getDescription() {
+        return description;
+    }
+
+
+
+
+    /**
+     * 根据value获取对应的PlateType
+     * @param className
+     * @return PlateType
+     */
+    public static PlateType fromClassName(String className) {
+        for (PlateType type : values()) {
+            if (type.className.equals(className)) {
+                return type;
+            }
+        }
+        return null;
+    }
+
+}

smartjavaai-ocr/src/main/java/cn/smartjavaai/ocr/factory/PlateModelFactory.java

@@ -0,0 +1,168 @@
+package cn.smartjavaai.ocr.factory;
+
+import cn.smartjavaai.common.config.Config;
+import cn.smartjavaai.ocr.config.PlateDetModelConfig;
+import cn.smartjavaai.ocr.config.PlateRecModelConfig;
+import cn.smartjavaai.ocr.config.TableStructureConfig;
+import cn.smartjavaai.ocr.enums.PlateDetModelEnum;
+import cn.smartjavaai.ocr.enums.PlateRecModelEnum;
+import cn.smartjavaai.ocr.enums.TableStructureModelEnum;
+import cn.smartjavaai.ocr.exception.OcrException;
+import cn.smartjavaai.ocr.model.plate.CRNNPlateRecModel;
+import cn.smartjavaai.ocr.model.plate.PlateDetModel;
+import cn.smartjavaai.ocr.model.plate.PlateRecModel;
+import cn.smartjavaai.ocr.model.plate.Yolov5PlateDetModel;
+import cn.smartjavaai.ocr.model.table.CommonTableStructureModel;
+import cn.smartjavaai.ocr.model.table.TableStructureModel;
+import lombok.extern.slf4j.Slf4j;
+
+import java.util.Map;
+import java.util.Objects;
+import java.util.concurrent.ConcurrentHashMap;
+
+/**
+ * 车牌识别模型工厂
+ * @author dwj
+ */
+@Slf4j
+public class PlateModelFactory {
+
+    // 使用 volatile 和双重检查锁定来确保线程安全的单例模式
+    private static volatile PlateModelFactory instance;
+
+    /**
+     * 模型缓存
+     */
+    private static final ConcurrentHashMap<PlateDetModelEnum, PlateDetModel> detModelMap = new ConcurrentHashMap<>();
+
+    /**
+     * 模型缓存
+     */
+    private static final ConcurrentHashMap<PlateRecModelEnum, PlateRecModel> recModelMap = new ConcurrentHashMap<>();
+
+
+    /**
+     * 模型注册表
+     */
+    private static final Map<PlateDetModelEnum, Class<? extends PlateDetModel>> detModelRegistry =
+            new ConcurrentHashMap<>();
+
+    /**
+     * 模型注册表
+     */
+    private static final Map<PlateRecModelEnum, Class<? extends PlateRecModel>> recModelRegistry =
+            new ConcurrentHashMap<>();
+
+
+    public static PlateModelFactory getInstance() {
+        if (instance == null) {
+            synchronized (PlateModelFactory.class) {
+                if (instance == null) {
+                    instance = new PlateModelFactory();
+                }
+            }
+        }
+        return instance;
+    }
+
+
+
+    /**
+     * 注册模型
+     * @param plateDetModelEnum
+     * @param clazz
+     */
+    private static void registerDetModel(PlateDetModelEnum plateDetModelEnum, Class<? extends PlateDetModel> clazz) {
+        detModelRegistry.put(plateDetModelEnum, clazz);
+    }
+
+    /**
+     * 注册模型
+     * @param plateRecModelEnum
+     * @param clazz
+     */
+    private static void registerRecModel(PlateRecModelEnum plateRecModelEnum, Class<? extends PlateRecModel> clazz) {
+        recModelRegistry.put(plateRecModelEnum, clazz);
+    }
+
+
+    /**
+     * 获取模型
+     * @param config
+     * @return
+     */
+    public PlateDetModel getDetModel(PlateDetModelConfig config) {
+        if(Objects.isNull(config) || Objects.isNull(config.getModelEnum())){
+            throw new OcrException("未配置OCR模型");
+        }
+        return detModelMap.computeIfAbsent(config.getModelEnum(), k -> {
+            return createDetModel(config);
+        });
+    }
+
+    /**
+     * 获取模型
+     * @param config
+     * @return
+     */
+    public PlateRecModel getRecModel(PlateRecModelConfig config) {
+        if(Objects.isNull(config) || Objects.isNull(config.getModelEnum())){
+            throw new OcrException("未配置OCR模型");
+        }
+        return recModelMap.computeIfAbsent(config.getModelEnum(), k -> {
+            return createRecModel(config);
+        });
+    }
+
+
+
+    /**
+     * 创建检测模型
+     * @param config
+     * @return
+     */
+    private PlateDetModel createDetModel(PlateDetModelConfig config) {
+        Class<?> clazz = detModelRegistry.get(config.getModelEnum());
+        if(clazz == null){
+            throw new OcrException("Unsupported model");
+        }
+        PlateDetModel model = null;
+        try {
+            model = (PlateDetModel) clazz.newInstance();
+        } catch (InstantiationException | IllegalAccessException e) {
+            throw new OcrException(e);
+        }
+        model.loadModel(config);
+        return model;
+    }
+
+    /**
+     * 创建识别模型
+     * @param config
+     * @return
+     */
+    private PlateRecModel createRecModel(PlateRecModelConfig config) {
+        Class<?> clazz = recModelRegistry.get(config.getModelEnum());
+        if(clazz == null){
+            throw new OcrException("Unsupported model");
+        }
+        PlateRecModel model = null;
+        try {
+            model = (PlateRecModel) clazz.newInstance();
+        } catch (InstantiationException | IllegalAccessException e) {
+            throw new OcrException(e);
+        }
+        model.loadModel(config);
+        return model;
+    }
+
+
+    // 初始化默认算法
+    static {
+        registerDetModel(PlateDetModelEnum.YOLOV5, Yolov5PlateDetModel.class);
+        registerDetModel(PlateDetModelEnum.YOLOV7, Yolov5PlateDetModel.class);
+        registerRecModel(PlateRecModelEnum.PLATE_REC_CRNN, CRNNPlateRecModel.class);
+        log.debug("缓存目录：{}", Config.getCachePath());
+    }
+
+}

smartjavaai-ocr/src/main/java/cn/smartjavaai/ocr/model/common/detect/OcrCommonDetModel.java

@@ -1,8 +1,11 @@
 package cn.smartjavaai.ocr.model.common.detect;
 
+import ai.djl.inference.Predictor;
 import ai.djl.modality.cv.Image;
+import ai.djl.ndarray.NDList;
 import cn.smartjavaai.ocr.config.OcrDetModelConfig;
 import cn.smartjavaai.ocr.entity.OcrBox;
+import org.apache.commons.pool2.impl.GenericObjectPool;
 
 import java.awt.image.BufferedImage;
 import java.util.List;
@@ -93,5 +96,9 @@ public interface OcrCommonDetModel extends AutoCloseable{
         throw new UnsupportedOperationException("默认不支持该功能");
     }
 
+    default GenericObjectPool<Predictor<Image, NDList>> getPool(){
+        throw new UnsupportedOperationException("默认不支持该功能");
+    }
+
 
 }

smartjavaai-ocr/src/main/java/cn/smartjavaai/ocr/model/common/detect/OcrCommonDetModelImpl.java

@@ -43,7 +43,7 @@ import java.util.*;
 @Slf4j
 public class OcrCommonDetModelImpl implements OcrCommonDetModel{
 
-    private ObjectPool<Predictor<Image, NDList>> detPredictorPool;
+    private GenericObjectPool<Predictor<Image, NDList>> detPredictorPool;
 
     private ZooModel<Image, NDList> detectionModel;
 
@@ -61,8 +61,14 @@ public class OcrCommonDetModelImpl implements OcrCommonDetModel{
             detectionModel = ModelZoo.loadModel(detCriteria);
             // 创建池子：每个线程独享 Predictor
             this.detPredictorPool = new GenericObjectPool<>(new PredictorFactory<>(detectionModel));
+            int predictorPoolSize = config.getPredictorPoolSize();
+            if(config.getPredictorPoolSize() <= 0){
+                predictorPoolSize = Runtime.getRuntime().availableProcessors(); // 默认等于CPU核心数
+            }
+            detPredictorPool.setMaxTotal(predictorPoolSize);
             log.debug("当前设备: " + detectionModel.getNDManager().getDevice());
             log.debug("当前引擎: " + Engine.getInstance().getEngineName());
+            log.debug("模型推理器线程池最大数量: " + predictorPoolSize);
         } catch (IOException | ModelNotFoundException | MalformedModelException e) {
             throw new OcrException("检测模型加载失败", e);
         }
@@ -209,6 +215,11 @@ public class OcrCommonDetModelImpl implements OcrCommonDetModel{
         }
     }
 
+    @Override
+    public GenericObjectPool<Predictor<Image, NDList>> getPool() {
+        return detPredictorPool;
+    }
+
     @Override
     public void close() throws Exception {
         try {

smartjavaai-ocr/src/main/java/cn/smartjavaai/ocr/model/common/detect/criteria/OcrCommonDetCriterialFactory.java

@@ -25,7 +25,7 @@ public class OcrCommonDetCriterialFactory {
     public static Criteria<Image, NDList> createCriteria(OcrDetModelConfig config) {
         Device device = null;
         if(!Objects.isNull(config.getDevice())){
-            device = config.getDevice() == DeviceEnum.CPU ? Device.cpu() : Device.gpu();
+            device = config.getDevice() == DeviceEnum.CPU ? Device.cpu() : Device.gpu(config.getGpuId());
         }
         Criteria<Image, NDList> criteria = null;
         ConcurrentHashMap params = new ConcurrentHashMap<String, String>();

smartjavaai-ocr/src/main/java/cn/smartjavaai/ocr/model/common/direction/OcrDirectionModel.java

@@ -9,6 +9,7 @@ import cn.smartjavaai.ocr.entity.OcrBox;
 import cn.smartjavaai.ocr.entity.OcrInfo;
 import cn.smartjavaai.ocr.entity.OcrItem;
 import cn.smartjavaai.ocr.model.common.detect.OcrCommonDetModel;
+import org.apache.commons.pool2.impl.GenericObjectPool;
 import org.opencv.core.Mat;
 
 import java.awt.image.BufferedImage;
@@ -106,4 +107,8 @@ public interface OcrDirectionModel extends AutoCloseable{
         throw new UnsupportedOperationException("默认不支持该功能");
     }
 
+    default GenericObjectPool<Predictor<Image, DirectionInfo>> getPool() {
+        throw new UnsupportedOperationException("默认不支持该功能");
+    }
+
 }

smartjavaai-ocr/src/main/java/cn/smartjavaai/ocr/model/common/direction/PPOCRMobileV2ClsModel.java

@@ -55,7 +55,7 @@ import java.util.concurrent.ConcurrentHashMap;
 public class PPOCRMobileV2ClsModel implements OcrDirectionModel {
 
 
-    private ObjectPool<Predictor<Image, DirectionInfo>> predictorPool;
+    private GenericObjectPool<Predictor<Image, DirectionInfo>> predictorPool;
 
     private DirectionModelConfig config;
 
@@ -71,10 +71,6 @@ public class PPOCRMobileV2ClsModel implements OcrDirectionModel {
         }
         this.config = config;
         this.textDetModel = config.getTextDetModel();
-        Device device = null;
-        if(!Objects.isNull(config.getDevice())){
-            device = config.getDevice() == DeviceEnum.CPU ? Device.cpu() : Device.gpu();
-        }
         ConcurrentHashMap params = new ConcurrentHashMap<String, String>();
         if(StringUtils.isNotBlank(config.getBatchifier())){
             params.put("batchifier", config.getBatchifier());
@@ -84,8 +80,14 @@ public class PPOCRMobileV2ClsModel implements OcrDirectionModel {
             model = ModelZoo.loadModel(criteria);
             // 创建池子：每个线程独享 Predictor
             this.predictorPool = new GenericObjectPool<>(new PredictorFactory<>(model));
+            int predictorPoolSize = config.getPredictorPoolSize();
+            if(config.getPredictorPoolSize() <= 0){
+                predictorPoolSize = Runtime.getRuntime().availableProcessors(); // 默认等于CPU核心数
+            }
+            predictorPool.setMaxTotal(predictorPoolSize);
             log.debug("当前设备: " + model.getNDManager().getDevice());
             log.debug("当前引擎: " + Engine.getInstance().getEngineName());
+            log.debug("模型推理器线程池最大数量: " + predictorPoolSize);
         } catch (IOException | ModelNotFoundException | MalformedModelException e) {
             throw new OcrException("模型加载失败", e);
         }
@@ -369,6 +371,11 @@ public class PPOCRMobileV2ClsModel implements OcrDirectionModel {
         return textDetModel;
     }
 
+    @Override
+    public GenericObjectPool<Predictor<Image, DirectionInfo>> getPool() {
+        return predictorPool;
+    }
+
     @Override
     public void close() throws Exception {
         try {

smartjavaai-ocr/src/main/java/cn/smartjavaai/ocr/model/common/direction/criteria/DirectionCriteriaFactory.java

@@ -28,7 +28,7 @@ public class DirectionCriteriaFactory {
     public static Criteria<Image, DirectionInfo> createCriteria(DirectionModelConfig config) {
         Device device = null;
         if(!Objects.isNull(config.getDevice())){
-            device = config.getDevice() == DeviceEnum.CPU ? Device.cpu() : Device.gpu();
+            device = config.getDevice() == DeviceEnum.CPU ? Device.cpu() : Device.gpu(config.getGpuId());
         }
         Criteria<Image, DirectionInfo> criteria = null;
         ConcurrentHashMap params = new ConcurrentHashMap<String, String>();

smartjavaai-ocr/src/main/java/cn/smartjavaai/ocr/model/common/recognize/OcrCommonRecModel.java

@@ -1,5 +1,6 @@
 package cn.smartjavaai.ocr.model.common.recognize;
 
+import ai.djl.inference.Predictor;
 import ai.djl.modality.cv.Image;
 import cn.smartjavaai.ocr.config.OcrDetModelConfig;
 import cn.smartjavaai.ocr.config.OcrRecModelConfig;
@@ -8,6 +9,7 @@ import cn.smartjavaai.ocr.entity.OcrBox;
 import cn.smartjavaai.ocr.entity.OcrInfo;
 import cn.smartjavaai.ocr.model.common.detect.OcrCommonDetModel;
 import cn.smartjavaai.ocr.model.common.direction.OcrDirectionModel;
+import org.apache.commons.pool2.impl.GenericObjectPool;
 
 import java.awt.image.BufferedImage;
 import java.util.List;
@@ -106,4 +108,8 @@ public interface OcrCommonRecModel extends AutoCloseable{
         throw new UnsupportedOperationException("默认不支持该功能");
     }
 
+    default GenericObjectPool<Predictor<Image, String>> getPool() {
+        throw new UnsupportedOperationException("默认不支持该功能");
+    }
+
 }

smartjavaai-ocr/src/main/java/cn/smartjavaai/ocr/model/common/recognize/OcrCommonRecModelImpl.java

@@ -24,7 +24,6 @@ import cn.smartjavaai.ocr.exception.OcrException;
 import cn.smartjavaai.ocr.model.common.detect.OcrCommonDetModel;
 import cn.smartjavaai.ocr.model.common.direction.OcrDirectionModel;
 import cn.smartjavaai.ocr.model.common.recognize.criteria.OcrCommonRecCriterialFactory;
-import cn.smartjavaai.ocr.opencv.OcrOpenCVUtils;
 import cn.smartjavaai.ocr.utils.OcrUtils;
 import lombok.extern.slf4j.Slf4j;
 import org.apache.commons.collections.CollectionUtils;
@@ -49,7 +48,7 @@ import java.util.stream.Collectors;
 @Slf4j
 public class OcrCommonRecModelImpl implements OcrCommonRecModel {
 
-    private ObjectPool<Predictor<Image, String>> recPredictorPool;
+    private GenericObjectPool<Predictor<Image, String>> recPredictorPool;
 
     private OcrRecModelConfig config;
 
@@ -72,8 +71,14 @@ public class OcrCommonRecModelImpl implements OcrCommonRecModel {
         try{
             recognitionModel = ModelZoo.loadModel(recCriteria);
             this.recPredictorPool = new GenericObjectPool<>(new PredictorFactory<>(recognitionModel));
+            int predictorPoolSize = config.getPredictorPoolSize();
+            if(config.getPredictorPoolSize() <= 0){
+                predictorPoolSize = Runtime.getRuntime().availableProcessors(); // 默认等于CPU核心数
+            }
+            recPredictorPool.setMaxTotal(predictorPoolSize);
             log.debug("当前设备: " + recognitionModel.getNDManager().getDevice());
             log.debug("当前引擎: " + Engine.getInstance().getEngineName());
+            log.debug("模型推理器线程池最大数量: " + predictorPoolSize);
         } catch (IOException | ModelNotFoundException | MalformedModelException e) {
             throw new OcrException("识别模型加载失败", e);
         }
@@ -238,7 +243,7 @@ public class OcrCommonRecModelImpl implements OcrCommonRecModel {
                 throw new OcrException("未检测到文字");
             }
             Mat wrappedImage = (Mat) img.getWrappedImage();
-            BufferedImage bufferedImage = OcrOpenCVUtils.mat2Image(wrappedImage);
+            BufferedImage bufferedImage = OpenCVUtils.mat2Image(wrappedImage);
             OcrUtils.drawRectWithText(bufferedImage, ocrInfo, fontSize);
             ImageUtils.saveImage(bufferedImage, outputPath);
             wrappedImage.release();
@@ -443,6 +448,11 @@ public class OcrCommonRecModelImpl implements OcrCommonRecModel {
         return directionModel;
     }
 
+
+    public GenericObjectPool<Predictor<Image, String>> getRecPredictorPool() {
+        return recPredictorPool;
+    }
+
     @Override
     public void close() throws Exception {
         try {

smartjavaai-ocr/src/main/java/cn/smartjavaai/ocr/model/common/recognize/criteria/OcrCommonRecCriterialFactory.java

@@ -24,7 +24,7 @@ public class OcrCommonRecCriterialFactory {
     public static Criteria<Image, String> createCriteria(OcrRecModelConfig config) {
         Device device = null;
         if(!Objects.isNull(config.getDevice())){
-            device = config.getDevice() == DeviceEnum.CPU ? Device.cpu() : Device.gpu();
+            device = config.getDevice() == DeviceEnum.CPU ? Device.cpu() : Device.gpu(config.getGpuId());
         }
         Criteria<Image, String> criteria = null;
         ConcurrentHashMap params = new ConcurrentHashMap<String, String>();

smartjavaai-ocr/src/main/java/cn/smartjavaai/ocr/model/plate/CRNNPlateRecModel.java

@@ -0,0 +1,342 @@
+package cn.smartjavaai.ocr.model.plate;
+
+import ai.djl.MalformedModelException;
+import ai.djl.engine.Engine;
+import ai.djl.inference.Predictor;
+import ai.djl.modality.cv.Image;
+import ai.djl.modality.cv.ImageFactory;
+import ai.djl.modality.cv.output.BoundingBox;
+import ai.djl.modality.cv.output.DetectedObjects;
+import ai.djl.repository.zoo.Criteria;
+import ai.djl.repository.zoo.ModelNotFoundException;
+import ai.djl.repository.zoo.ModelZoo;
+import ai.djl.repository.zoo.ZooModel;
+import cn.hutool.core.lang.UUID;
+import cn.hutool.core.lang.generator.UUIDGenerator;
+import cn.smartjavaai.common.entity.DetectionRectangle;
+import cn.smartjavaai.common.entity.R;
+import cn.smartjavaai.common.pool.PredictorFactory;
+import cn.smartjavaai.common.utils.Base64ImageUtils;
+import cn.smartjavaai.common.utils.FileUtils;
+import cn.smartjavaai.common.utils.ImageUtils;
+import cn.smartjavaai.common.utils.OpenCVUtils;
+import cn.smartjavaai.ocr.config.PlateDetModelConfig;
+import cn.smartjavaai.ocr.config.PlateRecModelConfig;
+import cn.smartjavaai.ocr.entity.PlateInfo;
+import cn.smartjavaai.ocr.entity.PlateResult;
+import cn.smartjavaai.ocr.enums.PlateType;
+import cn.smartjavaai.ocr.exception.OcrException;
+import cn.smartjavaai.ocr.model.plate.criteria.PlateDetCriterialFactory;
+import cn.smartjavaai.ocr.model.plate.criteria.PlateRecCriterialFactory;
+import cn.smartjavaai.ocr.utils.OcrUtils;
+import lombok.extern.slf4j.Slf4j;
+import org.apache.commons.collections.CollectionUtils;
+import org.apache.commons.lang3.StringUtils;
+import org.apache.commons.pool2.ObjectPool;
+import org.apache.commons.pool2.impl.GenericObjectPool;
+import org.opencv.core.Core;
+import org.opencv.core.Mat;
+import org.opencv.core.Rect;
+import org.opencv.core.Size;
+import org.opencv.imgproc.Imgproc;
+
+import javax.imageio.ImageIO;
+import java.awt.*;
+import java.awt.image.BufferedImage;
+import java.io.*;
+import java.nio.file.Files;
+import java.nio.file.Path;
+import java.nio.file.Paths;
+import java.util.ArrayList;
+import java.util.Iterator;
+import java.util.List;
+import java.util.Objects;
+
+/**
+ * @author dwj
+ */
+@Slf4j
+public class CRNNPlateRecModel implements PlateRecModel{
+
+
+    private GenericObjectPool<Predictor<Image, PlateResult>> recPredictorPool;
+
+    private ZooModel<Image, PlateResult> recModel;
+
+    private PlateRecModelConfig config;
+
+    @Override
+    public void loadModel(PlateRecModelConfig config) {
+        if(StringUtils.isBlank(config.getModelPath())){
+            throw new OcrException("modelPath is null");
+        }
+        this.config = config;
+        //初始化 检测Criteria
+        Criteria<Image, PlateResult> detCriteria = PlateRecCriterialFactory.createCriteria(config);
+        try{
+            recModel = ModelZoo.loadModel(detCriteria);
+            // 创建池子：每个线程独享 Predictor
+            this.recPredictorPool = new GenericObjectPool<>(new PredictorFactory<>(recModel));
+            int predictorPoolSize = config.getPredictorPoolSize();
+            if(config.getPredictorPoolSize() <= 0){
+                predictorPoolSize = Runtime.getRuntime().availableProcessors(); // 默认等于CPU核心数
+            }
+            recPredictorPool.setMaxTotal(predictorPoolSize);
+            log.debug("当前设备: " + recModel.getNDManager().getDevice());
+            log.debug("当前引擎: " + Engine.getInstance().getEngineName());
+            log.debug("模型推理器线程池最大数量: " + predictorPoolSize);
+        } catch (IOException | ModelNotFoundException | MalformedModelException e) {
+            throw new OcrException("检测模型加载失败", e);
+        }
+    }
+
+    @Override
+    public R<List<PlateInfo>> recognize(String imagePath) {
+        if(!FileUtils.isFileExists(imagePath)){
+            return R.fail(R.Status.FILE_NOT_FOUND);
+        }
+        Image img = null;
+        try {
+            img = ImageFactory.getInstance().fromFile(Paths.get(imagePath));
+            R<List<PlateInfo>> plateResult = recognize(img);
+            return plateResult;
+        } catch (IOException e) {
+            throw new OcrException("无效的图片", e);
+        } finally {
+            ((Mat)img.getWrappedImage()).release();
+        }
+    }
+
+    @Override
+    public R<List<PlateInfo>> recognizeBase64(String base64Image) {
+        if(StringUtils.isBlank(base64Image)){
+            return R.fail(R.Status.INVALID_IMAGE);
+        }
+        byte[] imageData = Base64ImageUtils.base64ToImage(base64Image);
+        return recognize(imageData);
+    }
+
+    @Override
+    public R<List<PlateInfo>> recognize(BufferedImage image) {
+        if(!ImageUtils.isImageValid(image)){
+            return R.fail(R.Status.INVALID_IMAGE);
+        }
+        Image img = ImageFactory.getInstance().fromImage(OpenCVUtils.image2Mat(image));
+        R<List<PlateInfo>> plateResult = recognize(img);
+        ((Mat)img.getWrappedImage()).release();
+        return plateResult;
+    }
+
+    @Override
+    public R<List<PlateInfo>> recognize(byte[] imageData) {
+        if(Objects.isNull(imageData)){
+            return R.fail(R.Status.INVALID_IMAGE);
+        }
+        return recognize(new ByteArrayInputStream(imageData));
+    }
+
+    @Override
+    public R<List<PlateInfo>> recognize(Image image) {
+        if(Objects.isNull(config.getPlateDetModel())){
+            return R.fail(R.Status.PARAM_ERROR.getCode(), "未指定车牌检测模型");
+        }
+        DetectedObjects detectedObjects = config.getPlateDetModel().detect(image);
+        if(Objects.isNull(detectedObjects) || detectedObjects.getNumberOfObjects() == 0){
+            return R.fail(R.Status.NO_OBJECT_DETECTED);
+        }
+        List<PlateInfo> plateInfoList = OcrUtils.convertToPlateInfo(detectedObjects, image);
+        Predictor<Image, PlateResult> predictor = null;
+        try {
+            predictor = recPredictorPool.borrowObject();
+            for (PlateInfo plateInfo : plateInfoList){
+                DetectionRectangle detectionRectangle = plateInfo.getDetectionRectangle();
+//                Image subImage = image.getSubImage(detectionRectangle.getX(), detectionRectangle.getY(), detectionRectangle.getWidth(), detectionRectangle.getHeight());
+                //透视变换
+                Image subImage = OcrUtils.transformAndCrop((Mat)image.getWrappedImage(), plateInfo.getBox());
+                //双层车牌
+                if(plateInfo.getPlateType() == PlateType.DOUBLE){
+                    Mat mergeImage = getSplitMerge((Mat)subImage.getWrappedImage());
+                    subImage = ImageFactory.getInstance().fromImage(mergeImage);
+                }
+                PlateResult plateResult = predictor.predict(subImage);
+                if(Objects.nonNull(plateResult)){
+                    plateInfo.setPlateNumber(plateResult.getPlateNo());
+                    plateInfo.setPlateColor(plateResult.getPlateColor());
+                }
+            }
+           return R.ok(plateInfoList);
+        } catch (Exception e) {
+            throw new OcrException("车牌识别错误", e);
+        }finally {
+            if (predictor != null) {
+                try {
+                    recPredictorPool.returnObject(predictor); //归还
+                } catch (Exception e) {
+                    log.warn("归还Predictor失败", e);
+                    try {
+                        predictor.close(); // 归还失败才销毁
+                    } catch (Exception ex) {
+                        log.error("关闭Predictor失败", ex);
+                    }
+                }
+            }
+        }
+    }
+
+    /**
+     * 双层车牌进行分割后识别
+     * @param img
+     * @return
+     */
+    private Mat getSplitMerge(Mat img) {
+        int h = img.rows();
+        int w = img.cols();
+
+        // 上半部分：高度的前 5/12
+        Rect upperRect = new Rect(0, 0, w, (int)(5.0 / 12 * h));
+        Mat imgUpper = new Mat(img, upperRect);
+
+        // 下半部分：高度从 1/3 开始
+        Rect lowerRect = new Rect(0, (int)(1.0 / 3 * h), w, h - (int)(1.0 / 3 * h));
+        Mat imgLower = new Mat(img, lowerRect);
+
+        // 将上半部分 resize 到与下半部分相同大小
+        Mat resizedUpper = new Mat();
+        Size lowerSize = imgLower.size();
+        Imgproc.resize(imgUpper, resizedUpper, lowerSize);
+
+        // 水平拼接（将上下拼成左右）
+        List<Mat> mergeList = new ArrayList<>();
+        mergeList.add(resizedUpper);
+        mergeList.add(imgLower);
+
+        Mat merged = new Mat();
+        Core.hconcat(mergeList, merged);
+        return merged;
+    }
+
+    @Override
+    public PlateResult recognizeCropped(Image image) {
+        Predictor<Image, PlateResult> predictor = null;
+        try {
+            predictor = recPredictorPool.borrowObject();
+            return predictor.predict(image);
+        } catch (Exception e) {
+            throw new OcrException("车牌检测错误", e);
+        }finally {
+            if (predictor != null) {
+                try {
+                    recPredictorPool.returnObject(predictor); //归还
+                } catch (Exception e) {
+                    log.warn("归还Predictor失败", e);
+                    try {
+                        predictor.close(); // 归还失败才销毁
+                    } catch (Exception ex) {
+                        log.error("关闭Predictor失败", ex);
+                    }
+                }
+            }
+
+        }
+    }
+
+    @Override
+    public R<List<PlateInfo>> recognize(InputStream inputStream) {
+        if(Objects.isNull(inputStream)){
+            return R.fail(R.Status.INVALID_IMAGE);
+        }
+        Image img = null;
+        try {
+            img = ImageFactory.getInstance().fromInputStream(inputStream);
+            return recognize(img);
+        } catch (IOException e) {
+            throw new OcrException("无效图片输入流", e);
+        } finally {
+            if (img != null){
+                ((Mat)img.getWrappedImage()).release();
+            }
+        }
+    }
+
+    @Override
+    public R<Void> recognizeAndDraw(String imagePath, String outputPath) {
+        if(!FileUtils.isFileExists(imagePath)){
+            return R.fail(R.Status.FILE_NOT_FOUND);
+        }
+        Image img = null;
+        try {
+            img = ImageFactory.getInstance().fromFile(Paths.get(imagePath));
+            R<List<PlateInfo>> plateResult = recognize(img);
+            if(!plateResult.isSuccess()){
+                return R.fail(plateResult.getCode(), plateResult.getMessage());
+            }
+            if(CollectionUtils.isEmpty(plateResult.getData())){
+                return R.fail(R.Status.NO_OBJECT_DETECTED);
+            }
+            BufferedImage bufferedImage = OpenCVUtils.mat2Image((Mat)img.getWrappedImage());
+            OcrUtils.drawPlateInfo(bufferedImage, plateResult.getData());
+            ImageIO.write(bufferedImage, "jpg", new File(outputPath));
+            return R.ok();
+        } catch (IOException e) {
+            throw new OcrException(e);
+        } finally {
+            if (img != null){
+                ((Mat)img.getWrappedImage()).release();
+            }
+        }
+    }
+
+    @Override
+    public R<BufferedImage> recognizeAndDraw(BufferedImage sourceImage) {
+        if(!ImageUtils.isImageValid(sourceImage)){
+            return R.fail(R.Status.INVALID_IMAGE);
+        }
+        Image img = ImageFactory.getInstance().fromImage(OpenCVUtils.image2Mat(sourceImage));
+        try {
+            R<List<PlateInfo>> plateResult = recognize(img);
+            if(!plateResult.isSuccess()){
+                return R.fail(plateResult.getCode(), plateResult.getMessage());
+            }
+            if(CollectionUtils.isEmpty(plateResult.getData())){
+                return R.fail(R.Status.NO_OBJECT_DETECTED);
+            }
+            OcrUtils.drawPlateInfo((Mat)img.getWrappedImage(), plateResult.getData());
+            ByteArrayOutputStream outputStream = new ByteArrayOutputStream();
+            // 调用 save 方法将 Image 写入字节流
+            img.save(outputStream, "png");
+            // 将字节流转换为 BufferedImage
+            byte[] imageBytes = outputStream.toByteArray();
+            return R.ok(ImageIO.read(new ByteArrayInputStream(imageBytes)));
+        } catch (IOException e) {
+            throw new OcrException("导出图片失败", e);
+        } finally {
+            if (img != null){
+                ((Mat)img.getWrappedImage()).release();
+            }
+        }
+    }
+
+
+    @Override
+    public GenericObjectPool<Predictor<Image, PlateResult>> getPool() {
+        return recPredictorPool;
+    }
+
+    @Override
+    public void close() throws Exception {
+        try {
+            if (recPredictorPool != null) {
+                recPredictorPool.close();
+            }
+        } catch (Exception e) {
+            log.warn("关闭 predictorPool 失败", e);
+        }
+        try {
+            if (recModel != null) {
+                recModel.close();
+            }
+        } catch (Exception e) {
+            log.warn("关闭 model 失败", e);
+        }
+    }
+}