I would like to check if I can use MediaPipe (https://developers.google.com/mediapipe) to extract the skeleton and construct the annotation of the skeleton information of my custom RGB video dataset.
Then, input that .json file into the custom_2d_skeleton.py to construct .pkl file and conduct training.
I tried to use openpose to extract the skeleton, but it takes forever to install it. So just curious if I can use the MediaPipe.
If so, do you happen to know where to change in the config
I would like to check if I can use MediaPipe (https://developers.google.com/mediapipe) to extract the skeleton and construct the annotation of the skeleton information of my custom RGB video dataset. Then, input that .json file into the custom_2d_skeleton.py to construct .pkl file and conduct training.
I tried to use openpose to extract the skeleton, but it takes forever to install it. So just curious if I can use the MediaPipe. If so, do you happen to know where to change in the config