Fitness apps
Exercise form scoring
Compare a rep's joint trajectory to a correct template and flag form breakdown in real time.
MoveDeep AIMovement-Intelligence SDK
tom.touati@move-deep.comThe Movement Genius
We extract joint-level kinematics and gait parameters from ordinary video — the insight that today requires a motion-capture lab, delivered as an embeddable model for fitness, elderly-care, sports, and physiotherapy.
Request accessNo pricing · No sign-up · B2B pilot
gait · frame 042
From footage to biomechanics in one pass.
Pipeline- 01
Video in
Any phone or fixed camera. No capture suits, no markers.
- 02
Pose estimation
2D keypoints plus full-body 3D reconstruction per frame.
- 03
Biomechanical model
Temporal transformers resolve 3D joint-angle trajectories.
- 04
Insights
Gait parameters, form scores, and matched-norm deviations.
Four embedded use cases.
Elderly care
Fall-risk & mobility trends
Track gait variability and stepping speed over weeks to surface decline before a fall happens.
Sports clubs
Technique analysis
Quantify stride mechanics and asymmetry from club cam footage, no lab required.
Physiotherapy
Auto 10-meter walk test
Run a standardized gait assessment in seconds and archive it as a longitudinal record.
Science · Method
Pose estimation grounded in a peer-reviewed biomechanical standard.
Our 3D reconstruction is benchmarked against OpenCap, a markerless, instrument-free method for measuring human movement in everyday environments. Rather than proprietary black boxes, we anchor joint-angle estimation to a published, reproducible pipeline and train temporal transformers on synthetic data rendered from openly licensed motion-capture corpora — a commercially clean, fully owned data path.
- – Synthetic mocap rendering for supervised training
- – Batch, containerized inference · ~500–1,000 GPU-hours
- – Longitudinal, matched-norm movement database
Reference
Uhlrich SD, et al. OpenCap: Human movement dynamics from smartphone videos. PLOS Computational Biology. 2023.
Stanford Neuromuscular Biomechanics Lab. Markerless estimation of 3D kinematics and kinetics in everyday environments from two smartphone videos.
Vision
A normative movement database across ages and activity levels.
The SDK flags deviations from matched population norms, surfacing movement anomalies early and enabling longitudinal movement-quality insights. Over time, anonymized and consented motion becomes a reference atlas of how healthy people move at every age — turning a single capture into a durable clinical signal.