Research map / Graph learning / Knowledge and temporal graphs
From General to Specific: Informative Scene Graph Generation via Balance Adjustment
Yuyu Guo, Lianli Gao, Xuanhan Wang, Yuxuan Hu, Xing Xu, Xu Lu, Heng Tao Shen, Jingkuan Song
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 Balanced Predicate Learning (ablation) | Transformer · Visual Genome PredCls | mR@100: 17.5 -> 31.7 | Improved |
| add Balanced Predicate Learning (ablation) | MotifNet · Visual Genome PredCls | mR@100: 15.8 -> 29.1 | Improved |
| add Balanced Predicate Learning (ablation) | VCTree · Visual Genome SGDet | mR@100: 6.9 -> 14.4 | Improved |
| add Semantic Adjustment on top of BPL (ablation) | Transformer+BPL · Visual Genome PredCls | mR@100: 31.7 -> 34.2 | Improved |
| add Semantic Adjustment on top of BPL (ablation) | VCTree+BPL · Visual Genome SGCls | mR@100: 19.5 -> 21.2 | Improved |
| use frozen SA matrix instead of finetuning it (ablation) | Transformer+BPL+SA · Visual Genome PredCls | mR@100: 32.1 -> 34.2 | Improved |
| use SA matrix from confusion statistics instead of random init (ablation) | Transformer+BPL · Visual Genome PredCls | mR@100: 31.3 -> 32.1 | Improved |
| add identity matrix weight alpha=1.0 instead of 0.0 (ablation) | SA transition matrix · Visual Genome PredCls | mR@100: 27.3 -> 34.2 | Improved |
| build target domain from Visual Genome instead of Wikipedia (ablation) | BPL · Visual Genome PredCls | mR@100: 28.0 -> 34.2 | Improved |
| train on Wikipedia-based target domain instead of general domain (ablation) | Transformer · Visual Genome PredCls | mR@100: 17.5 -> 28.0 | Improved |
| use pretrained-model undersampling instead of random (ablation) | Transformer+BPL+SA · Visual Genome PredCls | mR@20: 26.7 -> 28.5 | Improved |
| use pretrained-model undersampling instead of random (ablation) | Transformer+BPL+SA · Visual Genome SGDet | mR@20: 11.4 -> 11.4 | No clear effect |
Closest papers by meaning
- Informative Scene Graph Generation via Debiasing (IJCV 2025 · similarity 0.94)
- Foundation Models Based Scene Graph Generation (TPAMI 2026 · similarity 0.91)
- Leveraging Predicate and Triplet Learning for Scene Graph Generation (CVPR 2024 · similarity 0.90)
- Scene Graph-Grounded Image Generation (AAAI 2025 · similarity 0.89)
- LLM4SGG: Large Language Models for Weakly Supervised Scene Graph Generation (CVPR 2024 · similarity 0.89)
- From Pixels to Graphs: Open-Vocabulary Scene Graph Generation with Vision-Language Models (CVPR 2024 · 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.
