Yolov5-face using ONNX Runtime on C++

Face landmarks example using ONNX Runtime C++, support dynamic batchsize.

Dependencies

• ONNX Runtime
• CMake 3.13.3
• OpenCV 4.6.0

ONNX Runtime Installation

bash scripts/pull_onnx_lib.sh

Opencv Installation: Recommend version 4.6.0

https://github.com/opencv/opencv

Download pretrained yolov5s-face

https://github.com/deepcam-cn/yolov5-face

Convert model Pytorch to ONNX

$ bash scripts/convert_onnx.sh

Usages

$ mkdir build & cd build
$ cmake ..
$ make
$ ./yoloface

References

Many thanks to these following projects

https://github.com/deepcam-cn/yolov5-face https://github.com/hpc203/yolov5-face-landmarks-opencv-v2