Research map / Image generation / Text-to-image diffusion
DEIG: Detail-Enhanced Instance Generation with Fine-Grained Semantic Control
Shiyan Du, Conghan Yue, Xinyu Cheng, Dongyu Zhang
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 Instance Detail Extractor (IDE) (ablation) | DEIG · DEIG-Bench (Qwen2.5-VL judge) | MAA human: 0.51 -> 0.75 | Improved |
| add Instance Detail Extractor (IDE) (ablation) | DEIG · DEIG-Bench (Qwen2.5-VL judge) | MAA object: 0.35 -> 0.44 | Improved |
| add Detail Fusion Module (masked attention) (ablation) | DEIG · DEIG-Bench (Qwen2.5-VL judge) | MAA human: 0.70 -> 0.75 | Improved |
| add Detail Fusion Module (masked attention) (ablation) | DEIG · DEIG-Bench (Qwen2.5-VL judge) | MAA object: 0.41 -> 0.44 | Improved |
| add detail-enriched instance captions for supervision (ablation) | DEIG · DEIG-Bench (Qwen2.5-VL judge) | MAA human: 0.31 -> 0.75 | Improved |
| add detail-enriched instance captions for supervision (ablation) | DEIG · DEIG-Bench (Qwen2.5-VL judge) | MAA object: 0.29 -> 0.44 | Improved |
| add detail-enriched instance captions for supervision (ablation) | DEIG · DEIG-Bench | mIoU: 0.70 -> 0.79 | Improved |
| add Instance Detail Extractor (IDE) (ablation) | DEIG · DEIG-Bench | mIoU: 0.73 -> 0.79 | Improved |
| increase aggregated semantic dimension S up to 16 (ablation) | DEIG · DEIG-Bench | MAA | Improved |
| increase aggregated semantic dimension S beyond 16 (ablation) | DEIG · DEIG-Bench | MAA | No clear effect |
| DEIG vs ROICtrl on human attribute binding | DEIG · DEIG-Bench (Qwen2.5-VL judge) | MAA human: 0.31 -> 0.75 | Improved |
| DEIG vs InstanceDiffusion on spatial precision | DEIG · InstDiff-Bench | AP: 0.40 -> 0.34 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- MIGC — MIGC: Multi-Instance Generation Controller for Text-to-Image Synthesis (CVPR 2024) · better on 7 of 7 shared results
- InstanceDiffusion — InstanceDiffusion: Instance-level Control for Image Generation (CVPR 2024) · better on 4 of 5 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
- ROICtrl: Boosting Instance Control for Visual Generation (CVPR 2025 · similarity 0.86)
- Localizing and Editing Knowledge In Text-to-Image Generative Models (ICLR 2024 · similarity 0.84)
- Patch Matters: Training-free Fine-grained Image Caption Enhancement via Local Perception (CVPR 2025 · similarity 0.84)
- Scene Graph Generation with Role-Playing Large Language Models (NeurIPS 2024 · similarity 0.83)
- Don't Just Chase “Highlighted Tokens” in MLLMs: Revisiting Visual Holistic Context Retention (NeurIPS 2025 · similarity 0.83)
- LLM Blueprint: Enabling Text-to-Image Generation with Complex and Detailed Prompts (ICLR 2024 · similarity 0.83)
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.
