Research map / 3D vision / Human pose estimation
EgoRenderer: Rendering Human Avatars From Egocentric Camera Images
Tao Hu, Kripasindhu Sarkar, Lingjie Liu, Matthias Zwicker, Christian Theobalt
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 |
|---|---|---|---|
| add the learned implicit texture stack (vs egocentric texture only) (ablation) | EgoRenderer · H1 single-video | LPIPS (x10): 1.769 -> 1.623 | Improved |
| add egocentric texture (vs implicit multi-view texture only) (ablation) | EgoRenderer · H1 single-video | LPIPS (x10): 1.738 -> 1.623 | Improved |
| learn the texture stack during training (vs static explicit texture) (ablation) | EgoRenderer · H3 single-video | LPIPS (x10): 1.676 -> 1.569 | Improved |
| use texture synthesis with implicit texture instead of direct pose-to-image translation (ablation) | Only-MV vs Pix2PixHD · H2 single-video | LPIPS (x10): 1.713 -> 1.626 | Improved |
| replace per-frame FeatureNet features with a global implicit texture stack (ablation) | EgoRenderer vs Fea-Net · H2 single-video | LPIPS (x10): 1.687 -> 1.617 | Improved |
| add the implicit texture stack (vs egocentric texture only) (ablation) | EgoRenderer · H4 single-video | SSIM (x10): 7.586 -> 7.535 | Hurt |
| add the implicit texture stack (vs static explicit texture), L1 distance (ablation) | EgoRenderer · single-video H1 and H3 | L1 distance (x10): 0.950 -> 0.994 (H1); 1.562 -> 1.448 (H3) | Mixed |
| train on multiple camera videos instead of a single video | Im-Tex · H1 | LPIPS (x10): 1.623 -> 1.174 | Improved |
| train on multiple camera videos instead of a single video | Im-Tex · H4 | LPIPS (x10): 1.571 -> 1.246 | Improved |
| train on multiple camera videos instead of a single video | Im-Tex · H1 | SSIM (x10): 7.529 -> 8.134 | Improved |
Closest papers by meaning
- SelfPose: 3D Egocentric Pose Estimation From a Headset Mounted Camera (TPAMI 2020 · similarity 0.90)
- xR-EgoPose: Egocentric 3D Human Pose From an HMD Camera (ICCV 2019 · similarity 0.90)
- Textured Neural Avatars (CVPR 2019 · similarity 0.90)
- FRAME: Floor-aligned Representation for Avatar Motion from Egocentric Video (CVPR 2025 · similarity 0.88)
- EgoGen: An Egocentric Synthetic Data Generator (CVPR 2024 · similarity 0.88)
- EgoSim: An Egocentric Multi-view Simulator and Real Dataset for Body-worn Cameras during Motion and Activity (NeurIPS 2024 · similarity 0.88)
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.
