CyCLeGen: Cycle-Consistent Layout Prediction and Image Generation in Vision Foundation Models
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866918390293069824 |
|---|---|
| author | Shan, Xiaojun Shen, Haoyu Mao, Yucheng Zhang, Xiang Anand, Abhay Li, Bingnan Xu, Haiyang Tu, Zhuowen |
| author_facet | Shan, Xiaojun Shen, Haoyu Mao, Yucheng Zhang, Xiang Anand, Abhay Li, Bingnan Xu, Haiyang Tu, Zhuowen |
| contents | We present CyCLeGen, a unified vision-language foundation model capable of both image understanding and image generation within a single autoregressive framework. Unlike existing vision models that depend on separate modules for perception and synthesis, CyCLeGen adopts a fully integrated architecture that enforces cycle-consistent learning through image->layout->image and layout->image->layout generation loops. This unified formulation introduces two key advantages: introspection, enabling the model to reason about its own generations, and data efficiency, allowing self-improvement via synthetic supervision under a reinforcement learning objective guided by cycle consistency. Extensive experiments show that CyCLeGen achieves significant gains across diverse image understanding and generation benchmarks, highlighting the potential of unified vision-language foundation models. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_14957 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | CyCLeGen: Cycle-Consistent Layout Prediction and Image Generation in Vision Foundation Models Shan, Xiaojun Shen, Haoyu Mao, Yucheng Zhang, Xiang Anand, Abhay Li, Bingnan Xu, Haiyang Tu, Zhuowen Computer Vision and Pattern Recognition We present CyCLeGen, a unified vision-language foundation model capable of both image understanding and image generation within a single autoregressive framework. Unlike existing vision models that depend on separate modules for perception and synthesis, CyCLeGen adopts a fully integrated architecture that enforces cycle-consistent learning through image->layout->image and layout->image->layout generation loops. This unified formulation introduces two key advantages: introspection, enabling the model to reason about its own generations, and data efficiency, allowing self-improvement via synthetic supervision under a reinforcement learning objective guided by cycle consistency. Extensive experiments show that CyCLeGen achieves significant gains across diverse image understanding and generation benchmarks, highlighting the potential of unified vision-language foundation models. |
| title | CyCLeGen: Cycle-Consistent Layout Prediction and Image Generation in Vision Foundation Models |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2603.14957 |