Research map / Label-efficient and robust learning / Recognition and long-tail
Prototype-Based Embedding Network for Scene Graph Generation
Chaofan Zheng, Xinyu Lyu, Lianli Gao, Bo Dai, Jingkuan Song
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 Prototype-guided Learning (PL) (ablation) | PE-Net without PL or PR · Visual Genome PredCls | mR@100: 20.0 -> 25.4 | Improved |
| Add Prototype-guided Learning (PL) (ablation) | PE-Net without PL or PR · Visual Genome PredCls | R@100: 68.2 -> 70.1 | Improved |
| Add Prototype Regularization (PR) to PL (ablation) | PE-Net(P) · Visual Genome PredCls | mR@100: 25.4 -> 33.8 | Improved |
| Add Prototype Regularization (PR) to PL (ablation) | PE-Net(P) · Visual Genome PredCls | R@100: 70.1 -> 67.2 | Hurt |
| Use prototype-aligned semantic representations with PL and PR | Visual Genome scene graph generation · Visual Genome PredCls | mR@100: PE-Net 33.8; BGNN 32.9 | Improved |
| Use prototype-based relation modeling | Visual Genome PredCls · Visual Genome PredCls | IV-O: 9.73 -> 0.74 | Improved |
| Use prototype regularization to separate predicate prototypes | Visual Genome PredCls · Visual Genome PredCls | IIVR-R: 2.72 -> 1.67 | Improved |
| Use PE-Net rather than RU-Net | Open Images V6 · Open Images V6 | score_wtd: 43.5 -> 44.9 | Improved |
| Use PE-Net rather than RU-Net | Open Images V6 · Open Images V6 | R@50: 76.9 -> 76.5 | Hurt |
Closest papers by meaning
- Learning To Generate Language-Supervised and Open-Vocabulary Scene Graph Using Pre-Trained Visual-Semantic Space (CVPR 2023 · similarity 0.90)
- Contextual Translation Embedding for Visual Relationship Detection and Scene Graph Generation (TPAMI 2020 · similarity 0.89)
- Leveraging Predicate and Triplet Learning for Scene Graph Generation (CVPR 2024 · similarity 0.89)
- Multi-Prototype Space Learning for Commonsense-Based Scene Graph Generation (AAAI 2024 · similarity 0.87)
- DSGG: Dense Relation Transformer for an End-to-end Scene Graph Generation (CVPR 2024 · similarity 0.87)
- Scene Graph Generation with Role-Playing Large Language Models (NeurIPS 2024 · 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.
