Research map / 3D vision / Human pose estimation
Does Learning Specific Features for Related Parts Help Human Pose Estimation?
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Learn part-specific features for data-driven groups of related body parts instead of using fully shared features (ablation) | Part-based branching network · MPII validation, invisible parts (ten hard joints) | PCKh@0.5: 70.29 -> 71.59 | Helped |
| Use data-driven mutual-information grouping instead of handcrafted body-structure grouping (ablation) | Part-based branching network · MPII validation | PCKh@0.5 | Helped |
| Increase specific-feature depth from D=1 to D=2 (ablation) | Part-based branching network · MPII validation | PCKh@0.5 | Hurt |
| Increase the number of specific-feature channels (ablation) | Part-based branching network · MPII validation | PCKh@0.5 | Mixed |
| Use a deep shared residual-block baseline instead of specific-feature branches (ablation) | Eight-stack hourglass network · MPII validation, ten hard joints overall | PCKh@0.5: 87.74 -> 88.14 | Hurt |
| Use five part groups rather than two (ablation) | Part-based branching network · MPII validation | PCKh@0.5 | Helped |
| Use more than five part groups (ablation) | Part-based branching network · MPII validation | PCKh@0.5 | No clear effect |
| Use overlapping part groups instead of disjoint groups (ablation) | Part-based branching network · MPII validation | PCKh@0.5 | No clear effect |
| Propagate specific features rather than shared features between stacked PBNs (ablation) | Stacked part-based branching network · MPII validation, three-stack network | PCKh@0.5: 87.30% versus 87.21% | Hurt |
| Propagate specific features rather than shared features between stacked PBNs (ablation) | Stacked part-based branching network · MPII validation, eight-stack network | PCKh@0.5: 88.14% versus 87.68% | Hurt |
| Remove ankle localization tasks from the hourglass network (ablation) | Hourglass network · MPII validation, upper-body parts | PCKh@0.5: +0.30% | Helped |
| Remove ankle localization tasks from the hourglass network (ablation) | Hourglass network · MPII validation, lower-body parts | PCKh@0.5: 0.45% | Hurt |
Closest papers by meaning
- HiPoser: 3D Human Pose Estimation with Hierarchical Shared Learning at Parts-Level Using Inertial Measurement Units (AAAI 2025 · similarity 0.86)
- Optimizing Human Pose Estimation Through Focused Human and Joint Regions (AAAI 2025 · similarity 0.85)
- GRPose: Learning Graph Relations for Human Image Generation with Pose Priors (AAAI 2025 · similarity 0.85)
- Video-Based Human Pose Regression via Decoupled Space-Time Aggregation (CVPR 2024 · similarity 0.85)
- PoseBH: Prototypical Multi-Dataset Training Beyond Human Pose Estimation (CVPR 2025 · similarity 0.85)
- Can Generative Video Models Help Pose Estimation? (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.