Compositional Text-to-image Synthesis With Attention Map Control Of Diffusion Models

Wang Ruichen, Chen Zekang, Chen Chen, Ma Jian, Lu Haonan, Lin Xiaodong. Arxiv 2023

[Paper] [Code]    
Attention Mechanism Has Code Merging Model Architecture Prompting Reinforcement Learning Tools Training Techniques Transformer

Recent text-to-image (T2I) diffusion models show outstanding performance in generating high-quality images conditioned on textual prompts. However, they fail to semantically align the generated images with the prompts due to their limited compositional capabilities, leading to attribute leakage, entity leakage, and missing entities. In this paper, we propose a novel attention mask control strategy based on predicted object boxes to address these issues. In particular, we first train a BoxNet to predict a box for each entity that possesses the attribute specified in the prompt. Then, depending on the predicted boxes, a unique mask control is applied to the cross- and self-attention maps. Our approach produces a more semantically accurate synthesis by constraining the attention regions of each token in the prompt to the image. In addition, the proposed method is straightforward and effective and can be readily integrated into existing cross-attention-based T2I generators. We compare our approach to competing methods and demonstrate that it can faithfully convey the semantics of the original text to the generated content and achieve high availability as a ready-to-use plugin. Please refer to https://github.com/OPPOMente-Lab/attention-mask-control.

Similar Work