CMake also seems to have issues with building to support CMake, which the work here seems to fix - however, it targets a now depreciated version of nvidia/cuda. By changing the target from nvidia/cuda:11.4.0-cudnn8-devel-ubuntu18.04 to nvidia/cuda:11.3.1-cudnn8-devel-ubuntu18.04, it now has a functional source image.
The above with some removed dependencies for GPU support, namely the usage of caffe-cpu
instead of libcaffe-cuda-dev.
Pull it in the following manner:
docker pull ghcr.io/hiibolt/igait-openpose
Then, start it with the following command:
docker run -it --rm --gpus all --security-opt=label=disable ghcr.io/hiibolt/igait-openpose
Run an example inference:
./build/examples/openpose/openpose.bin --image_dir /openpose/examples/media --display 0 --write_images /output_images
Please note that in order for this to work, you will need to have the NVIDIA Container Toolkit installed.
- With the above example usage commands, a body pose skeleton should be mapped onto each output image
- You may expect a significantly slower experience compared to CUDA acceleration, which OpenPose will warn you of. If you intend to only use CPU, you may safely discard this error.
- Running
nvidia-smi
should display readily available GPUs. - Running an inference should debug output the availability and usage of at least one GPU
- With the above example command, a body pose skeleton should be mapped onto each output image
Developed by @hiibolt on GitHub