DreamText: High Fidelity Scene Text Synthesis

2024-05-23

AI Summary

Existing scene text synthesis methods suffer from severe character distortions, repetitions, and omissions in multi-style scenarios due to the lack of character-level supervision and insufficient font generalization capability of text encoders. To address these issues, we propose a high-fidelity scene text synthesis framework. First, we introduce a novel latent-space character-level attention reconstruction mechanism that dynamically corrects the generation process via character-aware masking. Second, we design a joint alternating optimization paradigm for the text encoder and diffusion-based generator to enable adaptive learning of font diversity. Third, we establish a cross-attention-driven mask generation module and an end-to-end jointly trained pipeline. Our method achieves significant improvements over state-of-the-art approaches across multiple benchmarks. Quantitative and qualitative evaluations consistently demonstrate substantial gains in character integrity, readability, and visual realism.

Technology Category

• Natural Language Processing: Generation
• Computer Vision: Diffusion Models for Vision
• Machine Learning: Deep Generative Models & Autoencoders

Application Category

• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Scene text synthesis involves rendering specified texts onto arbitrary images. Current methods typically formulate this task in an end-to-end manner but lack effective character-level guidance during training. Besides, their text encoders, pre-trained on a single font type, struggle to adapt to the diverse font styles encountered in practical applications. Consequently, these methods suffer from character distortion, repetition, and absence, particularly in polystylistic scenarios. To this end, this paper proposes DreamText for high-fidelity scene text synthesis. Our key idea is to reconstruct the diffusion training process, introducing more refined guidance tailored to this task, to expose and rectify the model's attention at the character level and strengthen its learning of text regions. This transformation poses a hybrid optimization challenge, involving both discrete and continuous variables. To effectively tackle this challenge, we employ a heuristic alternate optimization strategy. Meanwhile, we jointly train the text encoder and generator to comprehensively learn and utilize the diverse font present in the training dataset. This joint training is seamlessly integrated into the alternate optimization process, fostering a synergistic relationship between learning character embedding and re-estimating character attention. Specifically, in each step, we first encode potential character-generated position information from cross-attention maps into latent character masks. These masks are then utilized to update the representation of specific characters in the current step, which, in turn, enables the generator to correct the character's attention in the subsequent steps. Both qualitative and quantitative results demonstrate the superiority of our method to the state of the art.

Problem

• Improves character-level guidance in scene text synthesis.
• Addresses font style diversity in text encoder training.
• Reduces character distortion, repetition, and absence in polystylistic scenarios.

Innovation

• Reconstructs diffusion training for character-level guidance
• Employs heuristic alternate optimization strategy
• Jointly trains text encoder and generator