MoveDeep AI’s Markerless Biomechanics SDK

Field notes

The proposed workflow accepts footage from a phone or fixed camera, reconstructs 3D pose and joint-angle trajectories, then emits gait parameters, form scores, and deviations from matched norms. The page names four intended applications: rep scoring, fall-risk and mobility tracking, sports technique analysis, and an automated 10-meter walk test. It also describes synthetic-data training from openly licensed motion-capture corpora and estimates 500–1,000 GPU-hours for batch inference. The page identifies the project as pre-incorporation and says the SDK is not yet launched.