Research map / 3D vision / Human pose estimation
EgoBody3M: Egocentric Body Tracking on a VR Headset using a Diverse Dataset
Amy Zhao, Chengcheng Tang, Lezi Wang, Yijing Li, Mihika Dave, Lingling Tao, Christopher D. Twigg, Robert Y. Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use multi-view latent features instead of 2D keypoints for lifting (ablation) | egocentric body tracker · EgoBody3M test | overall MPJPE (cm): 17.5 -> 5.49 | Improved |
| add temporal model (TCN + LSTM) instead of per-frame MLP (ablation) | Latent + MLP · EgoBody3M test | overall MPJPE (cm): 5.49 -> 5.18 | Improved |
| add temporal model instead of per-frame MLP (ablation) | Latent + MLP · EgoBody3M test | overall MPJVE (cm/frame): 0.97 -> 0.54 | Improved |
| add temporal model instead of per-frame MLP (ablation) | Latent + MLP · EgoBody3M test, out-of-view wrists | MPJPE (cm): 14.1 -> 12.2 | Improved |
| train end-to-end instead of 2-stage with frozen encoder (ablation) | Latent + temporal · EgoBody3M test | overall MPJPE (cm): 6.54 -> 5.18 | Improved |
| use proposed architecture instead of UnrealEgo | egocentric body tracking · EgoBody3M test | overall MPJPE (cm): 7.41 -> 5.18 | Improved |
| use proposed architecture instead of UnrealEgo with ImageNet pretraining | egocentric body tracking · UnrealEgo dataset | overall MPJPE (cm): 7.91 -> 6.70 | Improved |
| add temporal model but 2-stage training, compared with per-frame E2E MLP (ablation) | Latent + temporal 2-stage · EgoBody3M test | overall MPJPE (cm): 5.49 -> 6.54 | Hurt |
Closest papers by meaning
- EgoSim: An Egocentric Multi-view Simulator and Real Dataset for Body-worn Cameras during Motion and Activity (NeurIPS 2024 · similarity 0.92)
- FRAME: Floor-aligned Representation for Avatar Motion from Egocentric Video (CVPR 2025 · similarity 0.91)
- Instance Tracking in 3D Scenes from Egocentric Videos (CVPR 2024 · similarity 0.89)
- EMHI: A Multimodal Egocentric Human Motion Dataset with HMD and Body-Worn IMUs (AAAI 2025 · similarity 0.89)
- EgoGen: An Egocentric Synthetic Data Generator (CVPR 2024 · similarity 0.89)
- Ego4o: Egocentric Human Motion Capture and Understanding from Multi-Modal Input (CVPR 2025 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
