Jas
Also need to modify the dataset info file https://github.com/open-mmlab/mmpose/blob/master/configs/_base_/datasets/coco.py
It is not fully supported yet.
But we have provided many pre-trained models for whole-body pose estimation. You may have a try!
For now, we do not have a demo code for VoxelPose. It is still in development.
It seems that the green 'pose' is correct, but there are some translation (shift).
@ly015 Could you offer some help with this?
> Thanks! BTW, we have already provided 133-kpt COCO-Wholebody models. You can run this model to obtain foot keypoints.
See https://github.com/open-mmlab/mmpose/issues/781
We will use horizontal-flip in data augmentation. When we flip the image, the left_eye of a person will become the right_eye in the image.
This line indicates that the `left_eye` and the `right_eye` are a swap/flip pair.