The repo is multi-pose estimation for CPU, GPUs are not needed. you can run it in the Intel CPU with WeCam or video file.
The code was tested on Ubuntu 16.04 & Win10, with Python 3.6 and PyTorch. Intel CPUs &OpenVINO are needed for testing.
- python >= 3.6
- pytorch
- opencv
- imutils
- numpy
- OpenVINO >= 2020 R1, following official instruction.
# USAGE
python multi_keypoint.py --xml /path/to/your/model/*.xml --bin /path/to/your/model/*.bin --input_video /path/to/your/video --output /path/to/save/video/file
tips: before running, please Set up the OpenVINO environment variables.
MIT License (refer to the LICENSE file for details).