Research map / 3D vision / Human pose estimation
From Sharp to Blur: Unsupervised Domain Adaptation for 2D Human Pose Estimation Under Extreme Motion Blur Using Event Cameras
Youngho Kim, Hoonhee Cho, Kuk-Jin Yoon
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 motion-aware event-based blur augmentation (MBA) (ablation) | Baseline student · EHPT-XC target blur | mAP: 37.5 -> 40.4 | Improved |
| add Stage 2 pseudo-label training (ablation) | Baseline student · EHPT-XC target blur | mAP: 37.5 -> 49.1 | Improved |
| add MBA on top of Stage 2 (ablation) | Ours · EHPT-XC target blur | mAP: 49.1 -> 50.4 | Improved |
| add Stages 3 and 4 refinement (ablation) | Ours · EHPT-XC target blur | mAP: 50.4 -> 51.6 | Improved |
| add motion augmentation (ablation) | Stage 1 teacher · EHPT-XC target blur | mAP: 46.5 -> 46.7 | Improved |
| add blur translation to motion augmentation (ablation) | Stage 1 teacher · EHPT-XC target blur | mAP: 46.7 -> 46.9 | Improved |
| combine image+event and event sub-teachers with refinement (ablation) | Stage 1 teacher · EHPT-XC target blur | mAP: 43.5 -> 46.9 | Improved |
| add refinement module to a single sub-teacher (ablation) | Stage 1 teacher · EHPT-XC target blur | mAP: 40.4 -> 39.6 | No clear effect |
| use refinement network instead of mean heatmap (ablation) | Stage 1 teacher · EHPT-XC target blur | mAP: 45.4 -> 46.9 | Improved |
| use refinement network instead of spatial attention (ablation) | Stage 1 teacher · EHPT-XC target blur | mAP: 41.9 -> 46.9 | Improved |
| set Stage 2 mask threshold th to 0.1 instead of 0.3 (ablation) | Ours · EHPT-XC target blur | mAP: 45.7 -> 50.4 | Improved |
| set Stage 4 mask threshold th' to 0.1 instead of 0.05 (ablation) | Ours · EHPT-XC target blur | mAP: 49.7 -> 51.6 | Improved |
Closest papers by meaning
- Event-based Motion Deblurring with Unpaired Data (CVPR 2026 · similarity 0.88)
- A Benchmark Dataset for Event-Guided Human Pose Estimation and Tracking in Extreme Conditions (NeurIPS 2024 · similarity 0.87)
- Continuous-Time Human Motion Field from Event Cameras (ICCV 2025 · similarity 0.87)
- Prior-guided Source-free Domain Adaptation for Human Pose Estimation (ICCV 2023 · similarity 0.87)
- Fast Adaptation for Human Pose Estimation via Meta-Optimization (CVPR 2024 · similarity 0.85)
- HumanMM: Global Human Motion Recovery from Multi-shot Videos (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.
