From 6948d710c20008fd3528c71e69822a2a2faa5d4f Mon Sep 17 00:00:00 2001 From: Jia Guo Date: Sat, 8 May 2021 20:45:20 +0800 Subject: [PATCH] Update README.md --- challenges/iccv21-mfr/README.md | 34 ++++++++++++++++++++++----------- 1 file changed, 23 insertions(+), 11 deletions(-) diff --git a/challenges/iccv21-mfr/README.md b/challenges/iccv21-mfr/README.md index 6713166..eab4d37 100644 --- a/challenges/iccv21-mfr/README.md +++ b/challenges/iccv21-mfr/README.md @@ -78,20 +78,32 @@ Inference time was evaluated on Tesla V100 GPU, using onnxruntime-gpu==1.6. ## Rules -1. We have two tracks, determined by the size of training dataset. - * Track A: Use MS1M-V3 as training set. - * Track B: Use Glint360K as training set. -2. Training set and testing set are both already aligned to 112x112, re-alignment is prohibited. -3. Mask data-augmentation is allowed, such as [this](https://github.com/deepinsight/insightface/tree/master/recognition/tools). The tool you used should be reproducible. +1. We have two tracks, determined by the size of training dataset and inference time limitation. + * Track A: Use MS1M-V3 as training set, download: [ref-link](https://github.com/deepinsight/insightface/tree/master/challenges/iccv19-lfr) + * Track B: Use Glint360K as training set, download: [ref-link](https://github.com/deepinsight/insightface/tree/master/recognition/partial_fc) +2. Training set and testing set are both aligned to 112x112, re-alignment is prohibited. +3. Mask data-augmentation is allowed, such as [this](https://github.com/deepinsight/insightface/tree/master/recognition/tools). The applied mask augmentation tool should be reproducible. 4. External dataset and pretrained models are both prohibited. -5. Participants submit onnx model, then get scores by online evaluation. Test set is invisible. +5. Participants submit onnx model, then get scores by our online evaluation. Test images are invisible. 6. Matching score is measured by cosine similarity. -7. Model size should be not larger than 1GB. -8. For Track A: feature length should be not larger than 512, and the inference time should be not larger than 10ms on Tesla V100 GPU. -9. For Track B: feature length should be not larger than 1024, and the inference time should be not larger than 20ms on Tesla V100 GPU. -10. The input size of submission model should be 112x112. +7. Model size should not be larger than 1GB. +8. For Track A: feature length should not be not larger than 512, and the inference time should not be larger than 10ms on Tesla V100 GPU. +9. For Track B: feature length should not be not larger than 1024, and the inference time should not be larger than 20ms on Tesla V100 GPU. +10. The input shape of submission model should equal to 3x112x112 (RGB order). 11. Online evaluation server uses onnxruntime-gpu==1.6, cuda==10.2, cudnn==8.0.5. +12. Any float-16 model weights is prohibited, as it will lead to incorrect model size estimiation. + +## Tutorial + +1. ArcFace-PyTorch (with Partial-FC), [link](https://github.com/deepinsight/insightface/tree/master/recognition/arcface_torch) +2. OneFlow, [link](https://github.com/deepinsight/insightface/tree/master/recognition/oneflow_face) +3. MXNet, [link](https://github.com/deepinsight/insightface/tree/master/recognition/ArcFace) ## Submission -Coming soon +1. Participants package the onnx model for submission using ``zip`` or ``tar -czf``. +2. Each participant can submit three times a day. +3. Please sign-up with the real organization name. You can hide the organization name in our system if you like. +4. You can decide which submission to be on the leaderboard by clicking the button. + +Link coming soon