EgoHDM (We are currently building a dataset, and once it's complete, we will release both the code and the dataset together for evaluation)

Paper | Project Page | Video

Official code release for the following paper: "EgoHDM: An Online Egocentric-Inertial Human Motion Capture, Localization, and Dense Mapping System"

We present an egocentric-inertial human motion capture system that simultaneously estimates a dense map of the scene, runs in near real-time, and is fast and robust to initialize. The system takes as input 6 body-worn IMUs and a head-worn RGB camera. It achieves unprecedented accuracy in terms of localization and mapping, and adapts better to non-flat terrain than previous work thanks to physics-based corrections leveraging a local elevation map.

TODOs

• Release the live demo vis and the data processing code.

• Release the full system code.

• Release evaluation code.

Installation 💽

Step 1: create conda environment