Guardado en:
| Autores principales: | , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | https://arxiv.org/abs/2508.16644 |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866908953012600832 |
|---|---|
| author | Mondal, Anindya Banerjee, Ayan Nag, Sauradip Llados, Josep Zhu, Xiatian Dutta, Anjan |
| author_facet | Mondal, Anindya Banerjee, Ayan Nag, Sauradip Llados, Josep Zhu, Xiatian Dutta, Anjan |
| contents | Diffusion models excel at photorealistic synthesis but struggle with precise object counts, especially in high-density settings. We introduce COUNTLOOP, a training-free framework that achieves precise instance control through iterative, structured feedback. Our method alternates between synthesis and evaluation: a VLM-based planner generates structured scene layouts, while a VLM-based critic provides explicit feedback on object counts, spatial arrangements, and visual quality to refine the layout iteratively. Instance-driven attention masking and cumulative attention composition further prevent semantic leakage, ensuring clear object separation even in densely occluded scenes. Evaluations on COCO-Count, T2I-CompBench, and two newly introduced high instance benchmarks show that COUNTLOOP reduces counting error by up to 57% and achieves the highest or comparable spatial quality scores across all benchmarks, while maintaining photorealism. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2508_16644 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | CountLoop: Training-Free High-Instance Image Generation via Iterative Agent Guidance Mondal, Anindya Banerjee, Ayan Nag, Sauradip Llados, Josep Zhu, Xiatian Dutta, Anjan Computer Vision and Pattern Recognition Diffusion models excel at photorealistic synthesis but struggle with precise object counts, especially in high-density settings. We introduce COUNTLOOP, a training-free framework that achieves precise instance control through iterative, structured feedback. Our method alternates between synthesis and evaluation: a VLM-based planner generates structured scene layouts, while a VLM-based critic provides explicit feedback on object counts, spatial arrangements, and visual quality to refine the layout iteratively. Instance-driven attention masking and cumulative attention composition further prevent semantic leakage, ensuring clear object separation even in densely occluded scenes. Evaluations on COCO-Count, T2I-CompBench, and two newly introduced high instance benchmarks show that COUNTLOOP reduces counting error by up to 57% and achieves the highest or comparable spatial quality scores across all benchmarks, while maintaining photorealism. |
| title | CountLoop: Training-Free High-Instance Image Generation via Iterative Agent Guidance |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2508.16644 |