MAVIS

Motion Analysis and Visual Interaction Suite turns webcam, video, or caller-supplied frames into hand landmarks.

Version 0.1.0 is available on PyPI. Start with the installation guide, then run mavis-track demo for a desktop preview without calibration. The studio also provides face, pose, and holistic preview modes.

For another application, use the integration guide: from mavis_track import HandTracker, then tracker.process(frame). Your project keeps control of capture and display.

MAVIS includes numeric recordings, timed replay, rectangular CSV export, optional UDP/LSL, and advanced calibrated cameras. Image-normalized landmarks, hand-centered model estimates, and calibrated world coordinates are distinct data products. Joint angles are geometric estimates.

See the CLI, troubleshooting, and release guide for further details.