Research map / Graph learning / Knowledge and temporal graphs
Factorizable Net: An Efficient Subgraph-based Framework for Scene Graph Generation
Yikang Li, Wanli Ouyang, Bolei Zhou, Jianping Shi, Chao Zhang, Xiaogang Wang
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 |
|---|---|---|---|
| use subgraph clustering instead of per-pair phrase features (64 boxes) (ablation) | F-Net · VG-MSDN | inference time (s/image): 0.65 -> 0.18 | Improved |
| use subgraph clustering instead of per-pair phrase features (64 boxes) (ablation) | F-Net · VG-MSDN | SGGen Rec@100: 10.81 -> 10.33 | Hurt |
| raise object proposals from 64 to 200 with subgraphs (ablation) | F-Net · VG-MSDN | SGGen Rec@100: 10.33 -> 12.02 | Improved |
| add 2-D feature map for subgraph features (ablation) | F-Net · VG-MSDN | SGGen Rec@50: 9.73 -> 9.90 | Improved |
| add one spatial-weighted message passing (SMP) structure (ablation) | F-Net · VG-MSDN | SGGen Rec@100: 12.08 -> 14.55 | Improved |
| add Spatial-sensitive Relation Inference (SRI) (ablation) | F-Net · VG-MSDN | SGGen Rec@100: 14.55 -> 16.12 | Improved |
| add Spatial-sensitive Relation Inference (SRI) (ablation) | F-Net · VG-MSDN | PhrDet Rec@50: 20.48 -> 22.54 | Improved |
| stack a second SMP module (1 -> 2) (ablation) | F-Net · VG-MSDN | SGGen Rec@100: 16.12 -> 16.47 | Improved |
| stack a second SMP module (1 -> 2) (ablation) | F-Net · VG-MSDN | inference time (s/image): 0.44 -> 0.55 | Hurt |
| compare F-Net 2-SMP against MSDN | F-Net · VG-MSDN | SGGen Rec@50: 10.72 -> 13.06 | Improved |
| compare F-Net 2-SMP against DR-Net | F-Net · VG-DR-Net | SGGen Rec@50: 20.79 -> 19.88 | Hurt |
| compare F-Net 2-SMP against DR-Net | F-Net · VG-DR-Net | PhrDet Rec@100: 27.57 -> 32.63 | Improved |
Closest papers by meaning
- Scene Graph Generation by Iterative Message Passing (CVPR 2017 · similarity 0.93)
- LinkNet: Relational Embedding for Scene Graph (NeurIPS 2018 · similarity 0.91)
- DSGG: Dense Relation Transformer for an End-to-end Scene Graph Generation (CVPR 2024 · similarity 0.91)
- Scene Graph Generation With External Knowledge and Image Reconstruction (CVPR 2019 · similarity 0.90)
- Scene Graph-Grounded Image Generation (AAAI 2025 · similarity 0.89)
- EGTR: Extracting Graph from Transformer for Scene Graph Generation (CVPR 2024 · similarity 0.88)
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.
