Research map / 3D vision / Human pose estimation
Human from Blur: Human Pose Tracking from Blurry Images
Yiming Zhao, Denys Rozumnyi, Jie Song, Otmar Hilliges, Marc Pollefeys, Martin R. Oswald
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 adversarial motion prior (AMP) (ablation) | HfB · BT-AMASS single-frame, blur rate [0.05, 0.1] | MPJPE (mm): 65.1 -> 56.3 | Improved |
| add adversarial motion prior (AMP) (ablation) | HfB · BT-AMASS single-frame, blur rate [0.4, 0.5] | MPJPE (mm): 98.5 -> 89.0 | Improved |
| add adversarial motion prior (AMP) (ablation) | HfB · BT-AMASS single-frame, blur rate [0.9, 1.1] | MPJPE (mm): 138.9 -> 140.5 | Hurt |
| add adversarial motion prior (AMP) (ablation) | HfB · BT-AMASS single-frame, blur rate [0.9, 1.1] | IoU: 0.645 -> 0.633 | Hurt |
| add adversarial motion prior (AMP) (ablation) | HfB · BT-AMASS single-frame, all blur rates | MPJPE (mm) | Mixed |
| refine METRO initialization with HfB test-time optimization | METRO · BT-AMASS single-frame, blur rate [0.05, 0.1] | MPJPE (mm): 70.0 -> 56.3 | Improved |
| refine METRO initialization with HfB test-time optimization | METRO · BT-AMASS single-frame, blur rate [0.9, 1.1] | MPJPE (mm): 146.4 -> 140.5 | Improved |
| replace SfB with HfB (human model) | SfB · BT-AMASS single-frame, blur rate [0.05, 0.1] | IoU: 0.498 -> 0.859 | Improved |
| replace MfB with HfB (two frames) | MfB · BT-AMASS two-frame, blur rate [0.9, 1.1] | IoU: 0.229 -> 0.656 | Improved |
| replace AfB + METRO with HfB at high blur | AfB + METRO · BC-CAPE single-frame, blur rate [0.5, 0.6] | MPJPE (mm): 133.3 -> 110.6 | Improved |
| replace Jin et al. + METRO with HfB at low blur | Jin et al. + METRO · BC-CAPE single-frame, blur rate <0.2 | MPJPE (mm): 77.2 -> 81.2 | Hurt |
| replace AfB (trained on dataset) with HfB at original blur | AfB + METRO · B-AIST++ original (blur rate 0.27) | PA-MPJPE (mm): 52.3 -> 69.1 | Hurt |
Closest papers by meaning
- ODHSR: Online Dense 3D Reconstruction of Humans and Scenes from Monocular Videos (CVPR 2025 · similarity 0.87)
- Motion-From-Blur: 3D Shape and Motion Estimation of Motion-Blurred Objects in Videos (CVPR 2022 · similarity 0.87)
- From Sharp to Blur: Unsupervised Domain Adaptation for 2D Human Pose Estimation Under Extreme Motion Blur Using Event Cameras (ICCV 2025 · similarity 0.87)
- Shape from Blur: Recovering Textured 3D Shape and Motion of Fast Moving Objects (NeurIPS 2021 · similarity 0.86)
- HumanMM: Global Human Motion Recovery from Multi-shot Videos (CVPR 2025 · similarity 0.86)
- Reconstructing People, Places, and Cameras (CVPR 2025 · similarity 0.85)
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.
