Research map / Graph learning / Knowledge and temporal graphs
PPDL: Predicate Probability Distribution Based Loss for Unbiased Scene Graph Generation
Wei Li, Haiwei Zhang, Qijie Bai, Guoqing Zhao, Ning Jiang, Xiaojie Yuan
Check 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 PPDL to VCTree | VCTree · VG150 PredCls | constrained mR@50: 17.9 -> 33.3 | Improved |
| Add PPDL to VCTree | VCTree · VG150 SGCls | constrained mR@50: 10.1 -> 21.8 | Improved |
| Add PPDL to VCTree | VCTree · VG150 SGGen | constrained mR@50: 6.9 -> 11.3 | Improved |
| Add PPDL to VCTree | VCTree · VG150 PredCls | constrained R@50: 66.4 -> 47.6 | Hurt |
| Add PPDL to VCTree | VCTree · VG150 SGGen | constrained R@50: 27.9 -> 20.1 | Hurt |
| Add PPDL to IMP | IMP · VG150 PredCls | constrained mR@50: 9.8 -> 24.8 | Improved |
| Add PPDL to MOTIFS | MOTIFS · VG150 PredCls | constrained mR@50: 14.0 -> 32.2 | Improved |
| Add PPDL to MOTIFS | MOTIFS · VG150 SGGen | constrained mR@100: 6.6 -> 13.5 | Improved |
| Add TDE and PPDL to VCTree | VCTree · VG150 PredCls | constrained mR@100: 28.7 -> 36.2 | Improved |
| Remove dynamic PPDM updating (ablation) | VCTree+PPDL · VG150 PredCls | constrained mR@100: 33.8 -> 32.3 | Hurt |
| Add the binarization function θ (ablation) | VCTree+PPDL · VG150 PredCls | constrained mR@100: 33.8 -> 33.8 | No clear effect |
| Set β to 0.01 instead of 0.03 (ablation) | VCTree+PPDL · VG150 PredCls | constrained mR@100: 33.6 -> 31.9 | Hurt |
Closest papers by meaning
- Learning to Generate an Unbiased Scene Graph by Using Attribute-Guided Predicate Features (AAAI 2023 · similarity 0.92)
- Panoptic Scene Graph Generation with Semantics-Prototype Learning (AAAI 2024 · similarity 0.91)
- Noise-Guided Predicate Representation Extraction and Diffusion-Enhanced Discretization for Scene Graph Generation (ICML 2025 · similarity 0.91)
- RA-SGG: Retrieval-Augmented Scene Graph Generation Framework via Multi-Prototype Learning (AAAI 2025 · similarity 0.89)
- Leveraging Predicate and Triplet Learning for 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.
