CycleNet：重新思考文本引导扩散模型中用于图像编辑的循环一致性
CycleNet: Rethinking Cycle Consistency in Text-Guided Diffusion for Image Manipulation
AI总结：
针对预训练扩散模型无配对图像到图像转换一致性不足的问题，提出引入循环一致性的CycleNet框架，在多粒度任务上表现更优，数据需求低、训练算力要求小，还新增了物体物理状态变化研究数据集。
摘要：
扩散模型（DMs）已在图像合成任务中取得突破，但缺乏用于一致性图像到图像（I2I）转换的直观交互接口。现有研究探索了多种方法来解决这一问题，包括基于掩码的方法、基于注意力的方法以及图像条件化方法。然而，在利用预训练DM实现无配对I2I转换的同时保持令人满意的一致性，仍然是一项关键挑战。本文提出CycleNet——一种将循环一致性引入DM以正则化图像编辑过程的新颖且简洁的方法。我们在不同粒度的无配对I2I任务上验证了CycleNet的性能。除场景和物体级转换外，我们还新增贡献了一个多域I2I转换数据集，用于研究物体的物理状态变化。实验结果表明，CycleNet在转换一致性和质量上表现更优，且仅需修改文本提示词即可为分布外场景生成高质量图像。CycleNet是一个实用框架，即便训练数据非常有限（约2k）也具备鲁棒性，且仅需最少的计算资源（1块GPU）即可完成训练。项目主页：https://cyclenetweb.github.io/
英文摘要：
Diffusion models (DMs) have enabled breakthroughs in image synthesis tasks but lack an intuitive interface for consistent image-to-image (I2I) translation. Various methods have been explored to address this issue, including mask-based methods, attention-based methods, and image-conditioning. However, it remains a critical challenge to enable unpaired I2I translation with pre-trained DMs while maintaining satisfying consistency. This paper introduces Cyclenet, a novel but simple method that incorporates cycle consistency into DMs to regularize image manipulation. We validate Cyclenet on unpaired I2I tasks of different granularities. Besides the scene and object level translation, we additionally contribute a multi-domain I2I translation dataset to study the physical state changes of objects. Our empirical studies show that Cyclenet is superior in translation consistency and quality, and can generate high-quality images for out-of-domain distributions with a simple change of the textual prompt. Cyclenet is a practical framework, which is robust even with very limited training data (around 2k) and requires minimal computational resources (1 GPU) to train. Project homepage: https://cyclenetweb.github.io/
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
