Research map / Graph learning / Knowledge and temporal graphs
RU-Net: Regularized Unrolling Network for Scene Graph Generation
Xin Lin, Changxing Ding, Jing Zhang, Yibing Zhan, Dacheng Tao
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 U-MP to baseline GMP without GDE (ablation) | RU-Net · VG SGCLS | R@100: 41.2 -> 43.1 | Improved |
| Add GDE to baseline GMP without U-MP (ablation) | RU-Net · VG PREDCLS | R@100: 67.8 -> 69.2 | Improved |
| Add both U-MP and GDE instead of using neither module (ablation) | RU-Net · VG SGCLS | R@100: 41.2 -> 43.3 | Improved |
| Add both U-MP and GDE instead of using neither module (ablation) | RU-Net · VG PREDCLS | R@100: 67.8 -> 69.6 | Improved |
| Set graph regularization p to 0.1 rather than p=2 (ablation) | RU-Net U-MP · VG SGCLS | R@100: 41.9 -> 43.3 | Improved |
| Increase U-MP layers from 2 to 5 (ablation) | RU-Net U-MP · VG SGCLS | R@100: 42.7 -> 43.3 | Improved |
| Set GDE weight τ to 0.1 rather than 0.2 (ablation) | RU-Net GDE · VG PREDCLS | R@100: 68.3 -> 69.6 | Improved |
| Group relationship predictions by object categories and remove groups with fewer than three elements, rather than regularizing predictions per image (ablation) | RU-Net GDE · VG PREDCLS | mR@100: 22.5 -> 24.2 | Improved |
| Use RU-Net instead of R-CAGCN◇ | VG PREDCLS · VG | mR@100: 19.9 -> 24.7 | Improved |
| Use RU-Net instead of RelDN | OI V4 · OI V4 | score_wtd: 44.6 -> 48.2 | Improved |
| Use RU-Net instead of RelDN | OI V6 · OI V6 | score_wtd: 40.8 -> 43.5 | Improved |
| Use RU-Net instead of Seq2Seq-RL | VRD · VRD relation detection | R@50: 26.1 -> 27.4 | Improved |
Later papers that compared with this method
1 later accepted paper reports RU-Net as a baseline in their results tables; 0 of them report a better number on most of the results they share with it.
- Devil's on the Edges: Selective Quad Attention for Scene Graph Generation (CVPR 2023 · better on 1 of 2 shared results)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- GPS-Net◇ — GPS-Net: Graph Property Sensing Network for Scene Graph Generation (CVPR 2020) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Unbiased Heterogeneous Scene Graph Generation with Relation-Aware Message Passing Neural Network (AAAI 2023 · similarity 0.90)
- TD²-Net: Toward Denoising and Debiasing for Video Scene Graph Generation (AAAI 2024 · similarity 0.88)
- Scene Graph Generation Strategy with Co-occurrence Knowledge and Learnable Term Frequency (ICML 2024 · similarity 0.87)
- DSGG: Dense Relation Transformer for an End-to-end Scene Graph Generation (CVPR 2024 · similarity 0.87)
- Kumaraswamy Wavelet for Heterophilic Scene Graph Generation (AAAI 2024 · similarity 0.87)
- Adaptive Self-training Framework for Fine-grained Scene Graph Generation (ICLR 2024 · similarity 0.86)
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.
