Research map / Image generation / Text-to-image diffusion
Scene Graph Guided Generation: Enable Accurate Relations Generation in Text-to-Image Models via Textural Rectification
Guibao Shen, Luozhou Wang, Jiantao Lin, Wenhang Ge, Chaozhe Zhang, Xin Tao, Di Zhang, Pengfei Wan, Guangyong Chen, Yijun Li, Ying-cong Chen
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 scene-graph triplet embedding through SG-Adapter after the CLIP text encoder | Stable Diffusion · MultiRels | SG-IoU: 0.157 -> 0.623 | Improved |
| add token-to-triplet SG mask to the adapter cross-attention (ablation) | SG-Adapter · MultiRels | SG-IoU: 0.316 -> 0.623 | Improved |
| fine-tune CLIP text encoder on the small dataset | Stable Diffusion · MS-COCO-Stuff val | FID: 25.0 -> 58.2 | Hurt |
| use LoRA adapter (no scene graph embedding) instead of SG-Adapter | Stable Diffusion · MultiRels | SG-IoU: 0.157 -> 0.145 | No clear effect |
| place scene-graph embedding in U-Net (GLIGEN adapter) instead of text encoder side | Stable Diffusion · MultiRels | SG-IoU: 0.157 -> 0.141 | No clear effect |
| train SG-Adapter on Flickr30k and compare with SG-to-image methods | SG-Adapter · COCO-Stuff val 256x256 | FID: 32.9 -> 25.1 | Improved |
| add SG-Adapter on image quality | Stable Diffusion · MS-COCO-Stuff val | FID: 25.0 -> 26.2 | Mixed |
Closest papers by meaning
- GraphDreamer: Compositional 3D Scene Synthesis from Scene Graphs (CVPR 2024 · similarity 0.90)
- Scene Graph Generation with Role-Playing Large Language Models (NeurIPS 2024 · similarity 0.89)
- Scene Graph Disentanglement and Composition for Generalizable Complex Image Generation (NeurIPS 2024 · similarity 0.88)
- Universal Scene Graph Generation (CVPR 2025 · similarity 0.88)
- R3CD: Scene Graph to Image Generation with Relation-Aware Compositional Contrastive Control Diffusion (AAAI 2024 · similarity 0.88)
- Scene Graph-Grounded Image Generation (AAAI 2025 · 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.
