Research map / 3D vision / Human pose estimation
Occlusion-Aware Siamese Network for Human Pose Estimation
Lu Zhou, Yingying Chen, Yunze Gao, Jinqiao Wang, Hanqing Lu
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 reconstruction without erasing (ablation) | 2-stack Hourglass · MPII val ablation | PCKh@0.5: 87.42 -> 87.76 | Improved |
| add erasing to reconstruction (E&R) (ablation) | 2-stack Hourglass · MPII val ablation | PCKh@0.5: 87.76 -> 87.91 | Improved |
| predict occlusion flag by classification instead of regression (ablation) | 2-stack Hourglass · MPII val ablation | PCKh@0.5: 87.44 -> 86.56 | Hurt |
| add feature mimicking loss (OT, non-detach) to E&R (ablation) | 2-stack Hourglass · MPII val ablation | PCKh@0.5: 88.09 -> 88.69 | Improved |
| use OT divergence instead of L2 mimicking (ablation) | 2-stack Hourglass · MPII val ablation | PCKh@0.5: 88.46 -> 88.69 | Improved |
| remove detach on un-occluded branch for OT mimicking (ablation) | 2-stack Hourglass · MPII val ablation | PCKh@0.5: 88.39 -> 88.69 | Improved |
| use cosine mimicking without detach vs with detach (ablation) | 2-stack Hourglass · MPII val ablation | PCKh@0.5: 88.23 -> 88.46 | Improved |
| replace siamese framework with distillation (ablation) | 2-stack Hourglass · MPII val ablation | PCKh@0.5 | No clear effect |
| add occlusion-aware siamese network to HRNet | HRNet-W32 · MPII val | PCKh@0.5: 90.33 -> 90.71 | Improved |
| add occlusion-aware siamese network to hourglass | 2-stack Hourglass · MPII val occluded joints | PCKh@0.5: 72.59 -> 74.31 | Improved |
| add occlusion-aware siamese network to hourglass | 2-stack Hourglass · MPII val un-occluded joints | PCKh@0.5: 92.06 -> 93.18 | Improved |
| add occlusion-aware siamese network to HRNet | HRNet · MPII val occluded joints | PCKh@0.5: 76.21 -> 77.72 | Improved |
Closest papers by meaning
- Occlusion-Aware Networks for 3D Human Pose Estimation in Video (ICCV 2019 · similarity 0.87)
- Human Pose Recognition via Occlusion-Preserving Abstract Images (ECCV 2024 · similarity 0.87)
- Occlude Them All: Occlusion-Aware Attention Network for Occluded Person Re-ID (ICCV 2021 · similarity 0.86)
- Peeking into occluded joints: A novel framework for crowd pose estimation (ECCV 2020 · similarity 0.86)
- Occ^2Net: Robust Image Matching Based on 3D Occupancy Estimation for Occluded Regions (ICCV 2023 · similarity 0.86)
- $\text{Di}^2\text{Pose}$: Discrete Diffusion Model for Occluded 3D Human Pose Estimation (NeurIPS 2024 · similarity 0.84)
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.
