Inference Yunet and SFace with GPU or TensorRT? about opencv_zoo

[fengyuentau]

1. Compile OpenCV with CUDA. Following is the CMake command I used, you will have to fit it in your case:

   cmake -D CMAKE_BUILD_TYPE=RELEASE \
    -D CMAKE_INSTALL_PREFIX=./install \
    -D BUILD_opencv_python2=OFF \
    -D BUILD_opencv_python3=ON \
    -D PYTHON3_LIBRARY=/usr/lib/aarch64-linux-gnu/libpython3.6m.so \
    -D PYTHON3_INCLUDE_DIR=/usr/include/python3.6m \
    -D PYTHON3_EXECUTABLE=/usr/bin/python3.6 \
    -D WITH_CUDA=ON \
    -D WITH_CUDNN=ON \
    -D OPENCV_DNN_CUDA=ON \
    -D CUDA_FAST_MATH=ON \
    -D ENABLE_FAST_MATH=1 \
    -D CUDA_ARCH_BIN=5.3 \
    -D WITH_CUBLAS=1 \
    -D CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-10.2 \
    -D OPENCV_EXTRA_MODULES_PATH=/home/opencv-cn/Workspace/opencv/opencv_contrib/modules  ..

   You can find your CUDA_ARCH_BIN at https://developer.nvidia.com/cuda-gpus (compute capability in the tables).
2. Initialize YuNet & SFace with CUDA backend and GPU target in demo scripts:

   yunet = YuNet(modelPath=args.model_yunet, backendId=cv.dnn.DNN_BACKEND_CUDA, targetId=cv.dnn.DNN_TARGET_CUDA)
   sface = SFace(modelPath=args.model_sface, backendId=backendId=cv.dnn.DNN_BACKEND_CUDA, targetId=cv.dnn.DNN_TARGET_CUDA)

And you can run demo scripts and perform inference of YuNet & SFace on GPU.

[hmh10098]

1. Compile OpenCV with CUDA. Following is the CMake command I used, you will have to fit it in your case:

   cmake -D CMAKE_BUILD_TYPE=RELEASE \
    -D CMAKE_INSTALL_PREFIX=./install \
    -D BUILD_opencv_python2=OFF \
    -D BUILD_opencv_python3=ON \
    -D PYTHON3_LIBRARY=/usr/lib/aarch64-linux-gnu/libpython3.6m.so \
    -D PYTHON3_INCLUDE_DIR=/usr/include/python3.6m \
    -D PYTHON3_EXECUTABLE=/usr/bin/python3.6 \
    -D WITH_CUDA=ON \
    -D WITH_CUDNN=ON \
    -D OPENCV_DNN_CUDA=ON \
    -D CUDA_FAST_MATH=ON \
    -D ENABLE_FAST_MATH=1 \
    -D CUDA_ARCH_BIN=5.3 \
    -D WITH_CUBLAS=1 \
    -D CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda-10.2 \
    -D OPENCV_EXTRA_MODULES_PATH=/home/opencv-cn/Workspace/opencv/opencv_contrib/modules  ..

   You can find your CUDA_ARCH_BIN at https://developer.nvidia.com/cuda-gpus (compute capability in the tables).

2. Initialize YuNet & SFace with CUDA backend and GPU target in demo scripts:

   yunet = YuNet(modelPath=args.model_yunet, backendId=cv.dnn.DNN_BACKEND_CUDA, targetId=cv.dnn.DNN_TARGET_CUDA)
   sface = SFace(modelPath=args.model_sface, backendId=backendId=cv.dnn.DNN_BACKEND_CUDA, targetId=cv.dnn.DNN_TARGET_CUDA)

And you can run demo scripts and perform inference of YuNet & SFace on GPU.

thanks for your help!