Research map / 3D vision / Human pose estimation
Correlating Edge, Pose With Parsing
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 |
|---|---|---|---|
| Use HNL to correlate parsing, edge, and pose features rather than the strongest listed baseline | Human parsing on LIP · LIP validation | mIoU: 2.23% mIoU | Helped |
| Use CorrPM rather than TGPNet | Human parsing on ATR · ATR test | F-1 score: 4.36% | Helped |
| Use CorrPM rather than Parsing R-CNN (X101) | Human parsing on CIHP · CIHP validation | mIoU: 0.38% mIoU | Helped |
| Add the edge branch to the parsing-only model through feature concatenation (ablation) | P on LIP · LIP validation | mIoU: 2.6% | Helped |
| Add the pose branch to the parsing-only model through feature concatenation (ablation) | P on LIP · LIP validation | mIoU: 2.12% | Helped |
| Concatenate both edge and pose features rather than edge features alone (ablation) | P+B on LIP · LIP validation | mIoU: 0.66% | Helped |
| Add parsing self-correlation (ablation) | P on LIP · LIP validation | mIoU: 0.09% mIoU | No clear effect |
| Correlate parsing with edge features rather than concatenate them (ablation) | P+B on LIP · LIP validation | mIoU: 1.84% | Helped |
| Correlate parsing with pose features rather than concatenate them (ablation) | P+K on LIP · LIP validation | mIoU: 3.19% | Helped |
| Use parsing-edge-pose correlation rather than concatenating all three features (ablation) | P+B+K on LIP · LIP validation | mIoU | Helped |
| Add direct parsing encoder loss L_p (ablation) | CorrPM on LIP · LIP validation | mIoU: 54.43 -> 55.33 | Helped |
| Keep the edge branch loss (α=2) rather than remove it (α=0), with β=70 (ablation) | CorrPM on LIP · LIP validation | mIoU: 53.98 -> 55.33 | Helped |
Closest papers by meaning
- Pose-Guided Human Parsing by an AND/OR Graph Using Pose-Context Features (AAAI 2016 · similarity 0.89)
- Devil in the Details: Towards Accurate Single and Multiple Human Parsing (AAAI 2019 · similarity 0.89)
- Look Into Person: Self-Supervised Structure-Sensitive Learning and a New Benchmark for Human Parsing (CVPR 2017 · similarity 0.88)
- CDGNet: Class Distribution Guided Network for Human Parsing (CVPR 2022 · similarity 0.87)
- Part-Aware Context Network for Human Parsing (CVPR 2020 · similarity 0.87)
- Weakly and Semi Supervised Human Body Part Parsing via Pose-Guided Knowledge Transfer (CVPR 2018 · similarity 0.86)
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.