【人脸识别】 新增多种人脸识别模型

【底层优化】 支持自由选择 OpenCV 或 BufferedImage 作为图像引擎

【通用图像】 全部模型启用 Image 输入,支持各类图片格式与 Image 的互转

【模型管理】 优化模型生命周期,关闭后可重新创建

【人脸识别】 支持在人脸查询结果中绘制姓名标注

【人脸检测】 新增人脸裁剪功能

【修复】 修复若干已知问题,提升系统稳定性
This commit is contained in:
dengwenjie
2025-10-02 16:26:42 +08:00
parent 1b50e2b943
commit dfa8cf9bb4
133 changed files with 6635 additions and 3532 deletions

View File

@@ -89,6 +89,7 @@ public class SpeechRecognizerFactory {
throw new AsrException(e);
}
model.loadModel(config);
model.setFromFactory(true);
return model;
}
@@ -100,4 +101,26 @@ public class SpeechRecognizerFactory {
log.debug("缓存目录:{}", Config.getCachePath());
}
/**
* 关闭所有已加载的模型
*/
public void closeAll() {
modelMap.values().forEach(model -> {
try {
model.close();
} catch (Exception e) {
e.printStackTrace();
}
});
modelMap.clear();
}
/**
* 移除缓存的模型
* @param modelEnum
*/
public static void removeFromCache(AsrModelEnum modelEnum) {
modelMap.remove(modelEnum);
}
}

View File

@@ -45,4 +45,7 @@ public interface SpeechRecognizer extends AutoCloseable{
throw new UnsupportedOperationException("默认不支持该功能");
}
default void setFromFactory(boolean fromFactory){
throw new UnsupportedOperationException("默认不支持该功能");
}
}

View File

@@ -9,6 +9,7 @@ import cn.smartjavaai.speech.asr.audio.SmartAudioFactory;
import cn.smartjavaai.speech.asr.config.AsrModelConfig;
import cn.smartjavaai.speech.asr.entity.*;
import cn.smartjavaai.speech.asr.exception.AsrException;
import cn.smartjavaai.speech.asr.factory.SpeechRecognizerFactory;
import cn.smartjavaai.speech.asr.pool.WhisperStatePool;
import cn.smartjavaai.speech.utils.AudioUtils;
import com.google.gson.JsonArray;
@@ -50,8 +51,11 @@ public class VoskRecognizer implements SpeechRecognizer{
private Model model;
private AsrModelConfig config;
@Override
public void loadModel(AsrModelConfig config) {
this.config = config;
if(StringUtils.isBlank(config.getModelPath())){
throw new AsrException("modelPath is null");
}
@@ -320,6 +324,9 @@ public class VoskRecognizer implements SpeechRecognizer{
@Override
public void close() throws Exception {
if (fromFactory) {
SpeechRecognizerFactory.removeFromCache(config.getModelEnum());
}
if(model != null){
model.close();
}
@@ -339,4 +346,13 @@ public class VoskRecognizer implements SpeechRecognizer{
}
private boolean fromFactory = false;
@Override
public void setFromFactory(boolean fromFactory) {
this.fromFactory = fromFactory;
}
public boolean isFromFactory() {
return fromFactory;
}
}

View File

@@ -12,6 +12,7 @@ import cn.smartjavaai.speech.asr.entity.AsrSegment;
import cn.smartjavaai.speech.asr.entity.RecParams;
import cn.smartjavaai.speech.asr.entity.WhisperParams;
import cn.smartjavaai.speech.asr.exception.AsrException;
import cn.smartjavaai.speech.asr.factory.SpeechRecognizerFactory;
import cn.smartjavaai.speech.asr.pool.WhisperStatePool;
import io.github.givimad.whisperjni.*;
import lombok.extern.slf4j.Slf4j;
@@ -41,8 +42,11 @@ public class WhisperRecognizer implements SpeechRecognizer{
private WhisperStatePool statePool;
private AsrModelConfig config;
@Override
public void loadModel(AsrModelConfig config) {
this.config = config;
if(StringUtils.isBlank(config.getModelPath())){
throw new AsrException("modelPath is null");
}
@@ -185,6 +189,9 @@ public class WhisperRecognizer implements SpeechRecognizer{
@Override
public void close() throws Exception {
if (fromFactory) {
SpeechRecognizerFactory.removeFromCache(config.getModelEnum());
}
if(statePool != null){
statePool.close();
}
@@ -216,4 +223,14 @@ public class WhisperRecognizer implements SpeechRecognizer{
false
);
}
private boolean fromFactory = false;
@Override
public void setFromFactory(boolean fromFactory) {
this.fromFactory = fromFactory;
}
public boolean isFromFactory() {
return fromFactory;
}
}