opencap-org/opencap-core

What it solves

OpenCap Core provides a way to estimate 3D human movement kinematics (joint angles) and 3D marker positions from two or more smartphone videos. This allows researchers to capture human movement dynamics without needing expensive, marker-based motion capture systems.

How it works

The system processes multiple video feeds to calculate 3D marker positions. It then uses these positions to estimate human movement kinematics in an OpenSim format. Users can run this pipeline in three ways: via a cloud-based web application, locally using data from the web app, or locally using near-synchronous videos from other sources.

Who it’s for

It is primarily designed for academic researchers studying human movement and biomechanics.

Highlights

• Multi-video input: Uses two or more videos to derive 3D data.
• OpenSim Integration: Outputs data in a format compatible with OpenSim for further kinematic and kinetic analysis.
• Flexible Deployment: Supports cloud processing via a web app or local execution for customization and high-accuracy pose estimation.
• Hardware Scalability: Local runs can be scaled from basic GPUs (4GB) to high-resolution settings requiring 24GB GPUs.