Research map / Graph learning / Knowledge and temporal graphs
Grapy-ML: Graph Pyramid Mutual Learning for Cross-Dataset Human Parsing
Read the paperCheck 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 |
|---|---|---|---|
| Add the graph pyramid module | DeepLab v3+ · CIHP | mIoU: 58.95 -> 60.36 | Helped |
| Add cross-dataset mutual learning with shared coarse levels | GPM · PASCAL-Person Part | mIoU: 69.50 -> 71.65 | Helped |
| Add cross-dataset mutual learning with shared coarse levels | GPM · ATR | mIoU: 76.97 -> 77.88 | Helped |
| Add cross-dataset mutual learning with shared coarse levels | GPM · CIHP | mIoU: 60.36 -> 60.60 | Helped |
| Replace no graph reasoning with graph convolutions (ablation) | DeepLab v3+ · CIHP | mIoU: 58.95 -> 59.81 | Helped |
| Use self-attention with average and max features for Level 3 reasoning instead of graph convolutions (ablation) | GPM · CIHP | mIoU: 59.81 -> 60.10 | Helped |
| Add Level 1 and Level 2 pyramid refinement to Level 3 average-and-max reasoning (ablation) | GPM · CIHP | mIoU: 60.10 -> 60.36 | Helped |
| Use average features only instead of both average and max features (ablation) | GPM · CIHP | mIoU: 60.36 -> 60.05 | Hurt |
| Use max features only instead of both average and max features (ablation) | GPM · CIHP | mIoU: 60.36 -> 59.89 | Hurt |
| Add mutual learning across datasets (ablation) | Full GPM · CIHP | mIoU: 60.36 -> 60.60 | Helped |
Closest papers by meaning
- Differentiable Multi-Granularity Human Representation Learning for Instance-Aware Human Semantic Parsing (CVPR 2021 · similarity 0.89)
- Graphonomy: Universal Human Parsing via Graph Transfer Learning (CVPR 2019 · similarity 0.89)
- Self-Learning With Rectification Strategy for Human Parsing (CVPR 2020 · similarity 0.88)
- Hierarchical Human Parsing With Typed Part-Relation Reasoning (CVPR 2020 · similarity 0.88)
- 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 0.88)
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.