Research map / Graph learning / Knowledge and temporal graphs
Can We Build Scene Graphs, Not Classify Them? FlowSG: Progressive Image-Conditioned Scene Graph Generation with Flow Matching
Xin Hu, Ke Qin, Wen Yin, Yuan-Fang Li, Ming Li, Tao He
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 flow-conditioned message aggregation (FMA) (ablation) | FlowSG · PSG closed-set SGDet | R@50: 40.5 -> 46.3 | Improved |
| add flow-conditioned message aggregation (FMA) (ablation) | FlowSG · PSG closed-set SGDet | mR@100: 39.7 -> 48.3 | Improved |
| add edge-level message aggregation (ablation) | FlowSG · PSG closed-set SGDet | R@50: 43.1 -> 46.3 | Improved |
| add node-level message aggregation (ablation) | FlowSG · PSG closed-set SGDet | mR@100: 42.5 -> 48.3 | Improved |
| add cross-attention to global visual features (ablation) | FlowSG · PSG closed-set SGDet | R@50: 39.2 -> 46.3 | Improved |
| increase codebook from 32x256 to 64x256 (ablation) | FlowSG · PSG closed-set SGDet | R@50: 32.7 -> 43.3 | Improved |
| increase codebook embedding from 64x256 to 64x512 (ablation) | FlowSG · PSG closed-set SGDet | R@50: 43.3 -> 46.3 | Improved |
| use 4 slots instead of 5 (ablation) | FlowSG · PSG closed-set SGDet | R@50: 35.6 -> 46.3 | Improved |
| use 4 slots instead of 3 (ablation) | FlowSG · PSG closed-set SGDet | R@50: 44.1 -> 46.3 | Improved |
| use 5 slots instead of 4 (ablation) | FlowSG · PSG closed-set SGDet | mR@100: 47.0 vs 48.3 | No clear effect |
| use marginal initialization of discrete states instead of uniform/masking/absorbing (ablation) | FlowSG · PSG closed-set SGDet | R/mR@50/100 | Improved |
Closest papers by meaning
- Adaptive Visual Scene Understanding: Incremental Scene Graph Generation (NeurIPS 2024 · similarity 0.91)
- DSGG: Dense Relation Transformer for an End-to-end Scene Graph Generation (CVPR 2024 · similarity 0.91)
- APT: Towards Universal Scene Graph Generation via Plug-in Adaptive Prompt Tuning (ICLR 2026 · similarity 0.91)
- Noise-Guided Predicate Representation Extraction and Diffusion-Enhanced Discretization for Scene Graph Generation (ICML 2025 · similarity 0.90)
- DiffVsgg: Diffusion-Driven Online Video Scene Graph Generation (CVPR 2025 · 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.
