HP-Edit: A Human-Preference Post-Training Framework for Image Editing
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866918459576680448 |
|---|---|
| author | Li, Fan Wang, Chonghuinan Lei, Lina Qiu, Yuping Xu, Jiaqi Jiang, Jiaxiu Qin, Xinran Chen, Zhikai Song, Fenglong Wang, Zhixin Pei, Renjing Zuo, Wangmeng |
| author_facet | Li, Fan Wang, Chonghuinan Lei, Lina Qiu, Yuping Xu, Jiaqi Jiang, Jiaxiu Qin, Xinran Chen, Zhikai Song, Fenglong Wang, Zhixin Pei, Renjing Zuo, Wangmeng |
| contents | Common image editing tasks typically adopt powerful generative diffusion models as the leading paradigm for real-world content editing. Meanwhile, although reinforcement learning (RL) methods such as Diffusion-DPO and Flow-GRPO have further improved generation quality, efficiently applying Reinforcement Learning from Human Feedback (RLHF) to diffusion-based editing remains largely unexplored, due to a lack of scalable human-preference datasets and frameworks tailored to diverse editing needs. To fill this gap, we propose HP-Edit, a post-training framework for Human Preference-aligned Editing, and introduce RealPref-50K, a real-world dataset across eight common tasks and balancing common object editing. Specifically, HP-Edit leverages a small amount of human-preference scoring data and a pretrained visual large language model (VLM) to develop HP-Scorer--an automatic, human preference-aligned evaluator. We then use HP-Scorer both to efficiently build a scalable preference dataset and to serve as the reward function for post-training the editing model. We also introduce RealPref-Bench, a benchmark for evaluating real-world editing performance. Extensive experiments demonstrate that our approach significantly enhances models such as Qwen-Image-Edit-2509, aligning their outputs more closely with human preference. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_19406 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | HP-Edit: A Human-Preference Post-Training Framework for Image Editing Li, Fan Wang, Chonghuinan Lei, Lina Qiu, Yuping Xu, Jiaqi Jiang, Jiaxiu Qin, Xinran Chen, Zhikai Song, Fenglong Wang, Zhixin Pei, Renjing Zuo, Wangmeng Computer Vision and Pattern Recognition Artificial Intelligence Common image editing tasks typically adopt powerful generative diffusion models as the leading paradigm for real-world content editing. Meanwhile, although reinforcement learning (RL) methods such as Diffusion-DPO and Flow-GRPO have further improved generation quality, efficiently applying Reinforcement Learning from Human Feedback (RLHF) to diffusion-based editing remains largely unexplored, due to a lack of scalable human-preference datasets and frameworks tailored to diverse editing needs. To fill this gap, we propose HP-Edit, a post-training framework for Human Preference-aligned Editing, and introduce RealPref-50K, a real-world dataset across eight common tasks and balancing common object editing. Specifically, HP-Edit leverages a small amount of human-preference scoring data and a pretrained visual large language model (VLM) to develop HP-Scorer--an automatic, human preference-aligned evaluator. We then use HP-Scorer both to efficiently build a scalable preference dataset and to serve as the reward function for post-training the editing model. We also introduce RealPref-Bench, a benchmark for evaluating real-world editing performance. Extensive experiments demonstrate that our approach significantly enhances models such as Qwen-Image-Edit-2509, aligning their outputs more closely with human preference. |
| title | HP-Edit: A Human-Preference Post-Training Framework for Image Editing |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence |
| url | https://arxiv.org/abs/2604.19406 |