RetinaFace

RetinaFace Network MXNet Symbol=>ONNX=>TensorRT

1.Reference

• RetinaFace arxiv: RetinaFace: Single-stage Dense Face Localisation in the Wild
• RetinaFace github: https://github.com/deepinsight/insightface/tree/master/detection/RetinaFace
• RetinaFaceAntiCov github: https://github.com/deepinsight/insightface/tree/master/detection/RetinaFaceAntiCov

2. Model Zoo

model	weights
RetinaFace-R50	weights
RetinaFace-MobileNet0.25	weights(nzof)
RetinaFaceAntiCov	weights(j3b6)

3.Export ONNX Model

• clone RetinaFace code

git clone https://github.com/deepinsight/insightface.git
cd insightface

copy export_onnx.py to ./detection/RetinaFace or ./detection/RetinaFaceAntiCov

• export resnet50 model

python3 export_onnx.py

• export mobilenet 0.25 model

python3 export_onnx.py  --prefix ./model/mnet.25

• export RetinaFaceAntiCov model

python3 export_onnx.py  --prefix ./model/mnet_cov2 --network net3l

4.Build tensorrt_inference Project

cd ../  # in project directory
mkdir build && cd build
cmake ..
make -j

5.Run tensorrt_inference

• inference with RetinaFace

cd ../../bin/
./tensorrt_inference RetinaFace ../configs/RetinaFace/config.yaml ../samples/faces_detection

• inference with RetinaFaceAntiCov

cd ../../bin/
./tensorrt_inference RetinaFace ../configs/RetinaFace/config_anti.yaml ../samples/faces_detection

For more information, please refer this blog: https://blog.csdn.net/linghu8812/article/details/110677016

6.Results

• RetinaFace result

• RetinaFaceAntiCov result