Research map / Graph learning / Knowledge and temporal graphs
GINet: Graph Interaction Network for Scene Parsing
Tianyi Wu, Yu Lu, Yu Zhu, Chuang Zhang, MingWu, Zhanyu Ma, Guodong Guo
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 visual-graph context reasoning to FCN+JPU baseline (ablation) | FCN+JPU baseline · PASCAL-Context (ablation, ResNet50) | mIoU (%): 48.5 -> 50.2 | Improved |
| add Graph Interaction unit (visual plus linguistic graph) to baseline (ablation) | FCN+JPU baseline · PASCAL-Context (ablation, ResNet50) | mIoU (%): 48.5 -> 51.0 | Improved |
| add linguistic semantic graph interaction over visual-graph-only reasoning (ablation) | GINet · PASCAL-Context (ablation, ResNet50) | mIoU (%): 50.2 -> 51.0 | Improved |
| add Semantic Context loss (ablation) | GINet · PASCAL-Context (ablation, ResNet50) | mIoU (%): 51.0 -> 51.7 | Improved |
| use ResNet-101 instead of ResNet50 backbone (ablation) | GINet · PASCAL-Context | mIoU (%): 51.7 -> 54.6 | Improved |
| use GINet instead of reproduced GCU | GINet · PASCAL-Context (ablation, ResNet50) | mIoU (%): 50.4 -> 51.7 | Improved |
| use GINet instead of reproduced GloRe | GINet · PASCAL-Context (ablation, ResNet50) | mIoU (%): 50.2 -> 51.7 | Improved |
| raise SC-loss weight above 0.2 (0.4 to 1.0) (ablation) | GINet · PASCAL-Context | mIoU (%) | No clear effect |
| swap GloVe for FastText or GoogleNews word embeddings (ablation) | GINet · PASCAL-Context | mIoU (%) | No clear effect |
| GINet versus DMNet on PASCAL-Context | GINet · PASCAL-Context | mIoU (%): 54.4 -> 54.9 | Improved |
| GINet versus ACNet (OHEM) on ADE20K | GINet · ADE20K val | mIoU (%): 45.90 -> 45.54 | Hurt |
| GINet versus ACNet on COCO Stuff | GINet · COCO Stuff test | mIoU (%): 40.1 -> 40.6 | Improved |
Closest papers by meaning
- Interaction via Bi-Directional Graph of Semantic Region Affinity for Scene Parsing (ICCV 2021 · similarity 0.90)
- Weakly Supervised Visual Semantic Parsing (CVPR 2020 · similarity 0.89)
- Graphical Contrastive Losses for Scene Graph Parsing (CVPR 2019 · similarity 0.89)
- Adaptive Context Network for Scene Parsing (ICCV 2019 · similarity 0.88)
- SCENIR: Visual Semantic Clarity through Unsupervised Scene Graph Retrieval (ICML 2025 · similarity 0.88)
- Scene Graph Thinking: Reinforcing Structured Visual Reasoning for Multimodal Large Language Models (ICML 2026 · 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.
