The goal is to do Real-time Human Pose Estimation with a range of algorithms in order to detect and track key human body joints. The ideal is to find options that work in every lighting condition, with noisy changing backgrounds, in realtime and with multi camera support. We've tested a variety of approaches including machine learning + RGB camera and machine learning + Depth camera.
Also see our Silhouette Segmentation Research .
https://ai.google.dev/edge/mediapipe/solutions/vision/pose_landmarker