Joint Low-level and High-level Textual Representation Learning with Multiple Masking Strategies

2025-05-11

AI Summary

To address performance degradation in text recognition caused by domain shift between synthetic and real-text images, this paper proposes a multi-mask strategy-driven self-supervised representation learning framework. The method introduces, for the first time, both blockwise and span masking into masked modeling of text images to explicitly capture character-level contextual relationships. Within a Masked Autoencoder (MAE) architecture, it jointly optimizes contrastive learning and masked image modeling (MIM) by synergistically fusing low-level texture and high-level semantic features. Extensive experiments demonstrate that the proposed approach significantly outperforms existing self-supervised methods on downstream tasks—including text recognition, text segmentation, and text image super-resolution—and achieves state-of-the-art (SOTA) performance across multiple real-world benchmarks.

Technology Category

• Machine Learning: Unsupervised & Self-Supervised Learning
• Computer Vision: Representation Learning for Vision
• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.

Application Category

• 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
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Most existing text recognition methods are trained on large-scale synthetic datasets due to the scarcity of labeled real-world datasets. Synthetic images, however, cannot faithfully reproduce real-world scenarios, such as uneven illumination, irregular layout, occlusion, and degradation, resulting in performance disparities when handling complex real-world images. Recent self-supervised learning techniques, notably contrastive learning and masked image modeling (MIM), narrow this domain gap by exploiting unlabeled real text images. This study first analyzes the original Masked AutoEncoder (MAE) and observes that random patch masking predominantly captures low-level textural features but misses high-level contextual representations. To fully exploit the high-level contextual representations, we introduce random blockwise and span masking in the text recognition task. These strategies can mask the continuous image patches and completely remove some characters, forcing the model to infer relationships among characters within a word. Our Multi-Masking Strategy (MMS) integrates random patch, blockwise, and span masking into the MIM frame, which jointly learns low and high-level textual representations. After fine-tuning with real data, MMS outperforms the state-of-the-art self-supervised methods in various text-related tasks, including text recognition, segmentation, and text-image super-resolution.

Problem

• Bridging performance gap between synthetic and real-world text images
• Enhancing high-level contextual learning in text recognition
• Integrating multiple masking strategies for improved textual representation

Innovation

• Introduces random blockwise and span masking
• Integrates multiple masking strategies in MIM
• Jointly learns low and high-level textual representations