Object-Attribute Binding in Text-to-Image Generation: Evaluation and Control

TL;DR

This paper introduces a new evaluation model and control method for text-to-image generation, improving attribute-object binding accuracy by leveraging syntactic constraints and disentangled embeddings.

Contribution

It proposes EPViT for evaluating image-text alignment and focused cross-attention (FCA) for better attribute-object binding without retraining diffusion models.

Findings

EPViT effectively evaluates image-text alignment.

FCA improves attribute-object binding in T2I generation.

Significant performance gains on multiple datasets.

Abstract

Current diffusion models create photorealistic images given a text prompt as input but struggle to correctly bind attributes mentioned in the text to the right objects in the image. This is evidenced by our novel image-graph alignment model called EPViT (Edge Prediction Vision Transformer) for the evaluation of image-text alignment. To alleviate the above problem, we propose focused cross-attention (FCA) that controls the visual attention maps by syntactic constraints found in the input sentence. Additionally, the syntax structure of the prompt helps to disentangle the multimodal CLIP embeddings that are commonly used in T2I generation. The resulting DisCLIP embeddings and FCA are easily integrated in state-of-the-art diffusion models without additional training of these models. We show substantial improvements in T2I generation and especially its attribute-object binding on several datasets.\footnote{Code and data will be made available upon acceptance.