Breaking

OpenCap Monocular estimates 3D gait from one phone video

July 10, 2026 at 19:53 EDT

Research & Papers
Healthcare
Open Source

OpenCap Monocular, a new version of the Stanford-originated motion analysis tool OpenCap, was released in beta in March 2026 as a method that estimates 3D human kinematics and musculoskeletal kinetics from a single smartphone video. From a clip under two minutes captured on one iPhone, it computes joint angles and positions (kinematics) as well as ground reaction forces, joint moments and muscle forces (kinetics), shipping alongside a research paper and a free cloud-processing beta.

March 25, 2026 · OpenCap Monocular

Lab-grade biomechanics from a single phone video

A new beta estimates 3D joint angles, ground reaction forces and muscle forces from one under-two-minute iPhone clip — no calibration, no neutral pose, no multi-camera rig.

smartphone needed — down from 2 iPhones + a PC

<2min

capture clip, processed free in the cloud

4.8°

mean error, rotational degrees of freedom

Accuracy gain vs. regression-only baseline

Validated against marker-based motion capture & force plates

+48%

Rotation
p = 0.036

+69%

Translation
p < 0.001

Pelvic translation error of just 3.4 cm — with walking ground reaction forces equal or better than the prior 2-camera OpenCap.

From video to musculoskeletal dynamics

Single phone clip

→

Monocular 3D pose + camera optimization

→

Physics + ML dynamics model

→

Kinematics + kinetics (OpenSim files)

What excites users

Near-lab accuracy for free, from a single phone and a roughly two-minute capture — scalable to clinical and at-home settings.

Open questions

Still a beta — not yet clinic-ready. TRC compatibility in the OpenSim flow, plus iOS-only support with no Android.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.