Research map / Detection and segmentation / CNN features
Bidirectional Graph Reasoning Network for Panoptic Segmentation
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 |
|---|---|---|---|
| Add BGRNet graph reasoning to Panoptic-FPN | Panoptic-FPN with ResNet50-FPN · COCO val | PQ: 42.5 -> 43.2 | Helped |
| Add BGRNet graph reasoning to Panoptic-FPN-D | Panoptic-FPN-D · ADE20K val | PQ: 30.1 -> 31.8 | Helped |
| Add an attention-based Thing-Graph (ablation) | Panoptic-FPN-D · ADE20K val | PQ: 30.1 -> 30.6 | Helped |
| Add a class-center Stuff-Graph (ablation) | Panoptic-FPN-D · ADE20K val | PQ: 30.1 -> 30.7 | Helped |
| Add both branch graphs without inter-branch interaction (ablation) | Panoptic-FPN-D · ADE20K val | PQ: 30.1 -> 31.1 | Helped |
| Use a fixed knowledge graph instead of attention for Thing-Graph construction (ablation) | Thing-Graph · ADE20K val | PQ: 30.1 -> 30.4 | Helped |
| Use attention instead of a fixed knowledge graph for Thing-Graph construction (ablation) | Thing-Graph · ADE20K val | PQ: 30.4 -> 30.6 | Helped |
| Use a pixel-level non-local graph instead of class-center nodes (ablation) | Stuff-Graph · ADE20K val | PQ: 30.7 -> 30.6 | No clear effect |
| Use pixel-level non-local nodes instead of class centers (ablation) | Stuff-Graph · ADE20K val | PQ^St: 26.2 -> 26.3 | No clear effect |
| Use semantic similarity instead of attention for graph connection (ablation) | Thing-Graph and Stuff-Graph · ADE20K val | PQ: 31.1 -> 31.5 | Helped |
| Use attention instead of semantic similarity for graph connection (ablation) | Thing-Graph and Stuff-Graph · ADE20K val | PQ: 31.5 -> 31.8 | Helped |
| Add Thing-to-Stuff graph connection only (ablation) | Thing-Graph and Stuff-Graph · ADE20K val | PQ: 31.1 -> 31.4 | Helped |
Closest papers by meaning
- Unifying Training and Inference for Panoptic Segmentation (CVPR 2020 · similarity 0.91)
- Auto-Panoptic: Cooperative Multi-Component Architecture Search for Panoptic Segmentation (NeurIPS 2020 · similarity 0.90)
- Attention-Guided Unified Network for Panoptic Segmentation (CVPR 2019 · similarity 0.90)
- SOGNet: Scene Overlap Graph Network for Panoptic Segmentation (AAAI 2020 · similarity 0.90)
- BANet: Bidirectional Aggregation Network With Occlusion Handling for Panoptic Segmentation (CVPR 2020 · similarity 0.90)
- Panoptic, Instance and Semantic Relations: A Relational Context Encoder To Enhance Panoptic Segmentation (CVPR 2022 · similarity 0.89)
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.