Research map / Graph learning / Knowledge and temporal graphs
Bridging Knowledge Graphs to Generate Scene Graphs
Alireza Zareian, Svebor Karaman, Shih-Fu Chang
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 the commonsense graph via bridging (vs plain classifier) (ablation) | GB-Net · Visual Genome PredCls (GC) | mR@50: 15.4 -> 18.2 | Improved |
| add the commonsense graph via bridging (vs plain classifier) (ablation) | GB-Net · Visual Genome PredCls (GC) | R@50: 62.5 -> 67.0 | Improved |
| add the commonsense graph via bridging (vs plain classifier) (ablation) | GB-Net · Visual Genome SGGen (GC) | mR@50: 5.5 -> 6.1 | Improved |
| add the commonsense graph via bridging (vs plain classifier) (ablation) | GB-Net · Visual Genome SGGen (GC) | R@50: 25.3 -> 26.4 | Improved |
| use 3 message passing steps instead of 1 (ablation) | GB-Net · Visual Genome PredCls (GC) | mR@50: 15.6 -> 18.2 | Improved |
| use 3 message passing steps instead of 1 (ablation) | GB-Net · Visual Genome SGGen (GC) | R@50: 24.9 -> 26.4 | Improved |
| use 3 message passing steps instead of 2 (ablation) | GB-Net · Visual Genome PredCls (GC) | R@50: 66.7 -> 67.0 | No clear effect |
| use 3 message passing steps instead of 2 (ablation) | GB-Net · Visual Genome PredCls (GC) | mR@50: 18.2 -> 18.2 | No clear effect |
| add class-balanced loss (beta) instead of plain cross-entropy (ablation) | GB-Net · Visual Genome PredCls (GC) | mR@50: 19.3 -> 22.1 | Improved |
| add class-balanced loss (beta) instead of plain cross-entropy (ablation) | GB-Net · Visual Genome SGCls (GC) | mR@50: 9.6 -> 12.7 | Improved |
| add class-balanced loss (beta) instead of plain cross-entropy (ablation) | GB-Net · Visual Genome (overall recall R) | R: -0.2 | No clear effect |
| use only one commonsense source (e.g. only ConceptNet) rather than all three (ablation) | GB-Net · Visual Genome | recall | No clear effect |
Closest papers by meaning
- Learning Visual Commonsense for Robust Scene Graph Generation (ECCV 2020 · similarity 0.93)
- SGEITL: Scene Graph Enhanced Image-Text Learning for Visual Commonsense Reasoning (AAAI 2022 · similarity 0.90)
- Scene Graph Generation by Iterative Message Passing (CVPR 2017 · similarity 0.90)
- A Comprehensive Survey of Scene Graphs: Generation and Application (TPAMI 2021 · similarity 0.90)
- Scene Graph Generation With External Knowledge and Image Reconstruction (CVPR 2019 · similarity 0.90)
- Scene Graph Thinking: Reinforcing Structured Visual Reasoning for Multimodal Large Language Models (ICML 2026 · 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.
