Research map / Graph learning / Knowledge and temporal graphs
Hierarchical Memory Learning for Fine-Grained Scene Graph Generation
Youming Deng, Yansheng Li, Yongjun Zhang, Xiang Xiang, Jian Wang, Jingdong Chen, Jiayi Ma
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 CR loss alone to MOTIFS (ablation) | MOTIFS · Visual Genome PredCls | mR@100: 17.19 -> 32.79 | Improved |
| add MR loss alone to MOTIFS (ablation) | MOTIFS · Visual Genome PredCls | mR@100: 17.19 -> 30.05 | Improved |
| add both CR and MR to MOTIFS (ablation) | MOTIFS · Visual Genome PredCls | mR@100: 32.79 -> 38.67 | Improved |
| add both CR and MR to Transformer (ablation) | Transformer · Visual Genome PredCls | mR@100: 19.38 -> 35.85 | Improved |
| add both CR and MR to VCTree (ablation) | VCTree · Visual Genome PredCls | mR@100: 18.08 -> 39.21 | Improved |
| add CR or MR (hierarchical training) to MOTIFS (ablation) | MOTIFS · Visual Genome PredCls | R@100: 67.20 -> 49.08 | Hurt |
| use 2 stages instead of 1 stage (ablation) | MOTIFS HML · Visual Genome PredCls | mR@100: 17.19 -> 38.67 | Improved |
| use 3 stages instead of 2 (ablation) | MOTIFS HML · Visual Genome PredCls | mR@100: 38.67 -> 34.44 | Hurt |
| use 4 stages instead of 2 (ablation) | MOTIFS HML · Visual Genome PredCls | mR@100: 38.67 -> 25.32 | Hurt |
| pass lambda 0.5 of information instead of none (ablation) | MOTIFS HML · Visual Genome PredCls | mR@100: 34.44 -> 38.67 | Improved |
| raise lambda from 0.5 to 1.00 (ablation) | MOTIFS HML · Visual Genome PredCls | mR@100: 38.67 -> 35.69 | Hurt |
| replace fine-grained predictions of on/has with general ones | MOTIFS HML · Visual Genome PredCls | R@100 of 'on': 31.01 -> 71.72 | Improved |
Closest papers by meaning
- RA-SGG: Retrieval-Augmented Scene Graph Generation Framework via Multi-Prototype Learning (AAAI 2025 · similarity 0.89)
- Adaptive Self-training Framework for Fine-grained Scene Graph Generation (ICLR 2024 · similarity 0.89)
- Leveraging Predicate and Triplet Learning for Scene Graph Generation (CVPR 2024 · similarity 0.89)
- Scene Graph Generation Strategy with Co-occurrence Knowledge and Learnable Term Frequency (ICML 2024 · similarity 0.88)
- Semi-Supervised Clustering Framework for Fine-grained Scene Graph Generation (AAAI 2025 · similarity 0.87)
- Noise-Guided Predicate Representation Extraction and Diffusion-Enhanced Discretization for Scene Graph Generation (ICML 2025 · similarity 0.87)
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.
