Research map / Image generation / GANs
Text2Scene: Generating Compositional Scenes From Textual Descriptions
Fuwen Tan, Song Feng, Vicente Ordonez
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 object and attribute attention instead of using no attention (ablation) | Text2Scene · Abstract Scenes test | U-obj precision: 0.665 -> 0.760 | Improved |
| Add dedicated attribute attention to object attention (ablation) | Text2Scene · Abstract Scenes test | U-obj precision: 0.731 -> 0.760 | Improved |
| Add attention regularization (ablation) | Text2Scene · Abstract Scenes test | U-obj precision: 0.749 -> 0.760 | Improved |
| Add object and attribute attention instead of using no attention (ablation) | Text2Scene · COCO val2017 | CIDEr: 0.531 -> 0.601 | Improved |
| Add dedicated attribute attention to object attention (ablation) | Text2Scene · COCO val2017 | CIDEr: 0.524 -> 0.601 | Improved |
| Add attention regularization (ablation) | Text2Scene · COCO val2017 | CIDEr: 0.555 -> 0.601 | Improved |
| Add regression-based inpainting post-processing (ablation) | Text2Scene synthetic composites · COCO val2014 | Inception Score: 22.33±1.58 -> 24.77±1.59 | Improved |
| Generate composite images with Text2Scene rather than competing methods | Text2Scene · COCO val2014 human comparison | Preference ratio over AttnGAN: 0.7588 | Improved |
Later papers that compared with this method
1 later accepted paper reports Text2Scene as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- A Parse-Then-Place Approach for Generating Graphic Layouts from Textual Descriptions (ICCV 2023 · better on 2 of 2 shared results)
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
- LLM Blueprint: Enabling Text-to-Image Generation with Complex and Detailed Prompts (ICLR 2024 · similarity 0.88)
- Compositional Scene Understanding through Inverse Generative Modeling (ICML 2025 · similarity 0.86)
- Scene Graph Generation with Role-Playing Large Language Models (NeurIPS 2024 · similarity 0.86)
- Progressive Compositionality in Text-to-Image Generative Models (ICLR 2025 · similarity 0.86)
- ConceptMix: A Compositional Image Generation Benchmark with Controllable Difficulty (NeurIPS 2024 · similarity 0.86)
- Enhancing Vision-Language Compositional Understanding with Multimodal Synthetic Data (CVPR 2025 · 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.
