Research map / Image generation / Text-to-image diffusion
CoLoGen: Progressive Learning of Concept-Localization Duality for Unified Image Generation
Yuxin Song, Yu Lu, Haoyuan Sun, Huanjin Yao, Fanglong Liu, Yifan Sun, Haocheng Feng, Hang Zhou, Jingdong 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 |
|---|---|---|---|
| add concept representation (Rc) experts (ablation) | CoLoGen · MagicBrush editing | CLIP-T: 0.260 -> 0.302 | Improved |
| add localization representation (Rl) (ablation) | CoLoGen · MagicBrush editing | CLIP-I: 0.889 -> 0.922 | Improved |
| add localization representation (Rl) (ablation) | CoLoGen · DreamBench | DINO: 0.683 -> 0.707 | Improved |
| add concept representation only (ablation) | CoLoGen · MagicBrush editing | CLIP-I: 0.889 -> 0.881 | Hurt |
| progressive staged training instead of multi-task co-training (ablation) | CoLoGen · MagicBrush editing | CLIP-T: 0.269 -> 0.301 | Improved |
| progressive staged training instead of multi-task co-training (ablation) | CoLoGen · DreamBench | DINO: 0.679 -> 0.714 | Improved |
| multi-task co-training of Rc and Rl vs baseline (ablation) | CoLoGen · DreamBench | DINO: 0.683 -> 0.679 | Hurt |
| add veteran gate routing supervision (ablation) | CoLoGen · MagicBrush editing (instruction-editing stage) | CLIP-I | Improved |
| routing density rho 0.8 vs other values (ablation) | CoLoGen · MagicBrush editing | CLIP-I | Improved |
| CoLoGen vs strongest controllable baselines on ADE20K segmentation | CoLoGen · Seg-to-Image (ADE20K) | mIoU: 44.23 -> 42.82 | Hurt |
| CoLoGen vs UniReal on editing l1 | CoLoGen · Emu Edit test | l1: 0.099 -> 0.111 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 4 on most shared results.
- Emu Edit — Emu Edit: Precise Image Editing via Recognition and Generation Tasks (CVPR 2024) · better on 3 of 4 shared results
- OmniGen — OmniGen: Unified Image Generation (CVPR 2025) · better on 1 of 3 shared results
- PixWizard — PixWizard: Versatile Image-to-Image Visual Assistant with Open-Language Instructions (ICLR 2025) · better on 1 of 1 shared result
- UniControl — UniControl: A Unified Diffusion Model for Controllable Visual Generation In the Wild (NeurIPS 2023) · better on 1 of 1 shared result
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
- On Mechanistic Knowledge Localization in Text-to-Image Generative Models (ICML 2024 · similarity 0.87)
- Bridging the gap to real-world language-grounded visual concept learning (NeurIPS 2025 · similarity 0.87)
- Localizing and Editing Knowledge In Text-to-Image Generative Models (ICLR 2024 · similarity 0.86)
- Easing Concept Bleeding in Diffusion via Entity Localization and Anchoring (ICML 2024 · similarity 0.86)
- Mechanisms of Object Localization in Vision-Language Models (CVPR 2026 · similarity 0.86)
- Concept Lancet: Image Editing with Compositional Representation Transplant (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.
