In this PR, I added a Dockerfile to build PyTorch and install Open-ReId (https://github.com/Cysu/open-reid).
Basically, we can use Open-ReID to extract the deep features for each human in the frame.
In reality, we are leverage the current OpenPose detection from pose_detection example to get the human bounding boxes. However, for simplicity, I've just extract the features for the entire frame.
More detailed information are in the Readme.md in this PR.
In this PR, I added a Dockerfile to build PyTorch and install Open-ReId (https://github.com/Cysu/open-reid). Basically, we can use Open-ReID to extract the deep features for each human in the frame.
In reality, we are leverage the current OpenPose detection from pose_detection example to get the human bounding boxes. However, for simplicity, I've just extract the features for the entire frame.
More detailed information are in the Readme.md in this PR.