Boosting Semi-Supervised Scene Text Recognition via Viewing and Summarizing

TL;DR

This paper introduces a novel semi-supervised scene text recognition framework inspired by viewing and summarizing human learning processes, utilizing contrastive learning, synthetic data augmentation, and a new alignment loss to improve recognition of challenging texts.

Contribution

It proposes a self-motivated contrastive learning framework with an online generation strategy and a character unidirectional alignment loss to enhance character morphology understanding and recognition accuracy.

Findings

Achieves state-of-the-art accuracy on benchmark datasets.

Effectively enriches character morphology diversity.

Improves recognition of complex and artistic texts.

Abstract

Existing scene text recognition (STR) methods struggle to recognize challenging texts, especially for artistic and severely distorted characters. The limitation lies in the insufficient exploration of character morphologies, including the monotonousness of widely used synthetic training data and the sensitivity of the model to character morphologies. To address these issues, inspired by the human learning process of viewing and summarizing, we facilitate the contrastive learning-based STR framework in a self-motivated manner by leveraging synthetic and real unlabeled data without any human cost. In the viewing process, to compensate for the simplicity of synthetic data and enrich character morphology diversity, we propose an Online Generation Strategy to generate background-free samples with diverse character styles. By excluding background noise distractions, the model is encouraged to focus on character morphology and generalize the ability to recognize complex samples when trained with only simple synthetic data. To boost the summarizing process, we theoretically demonstrate the derivation error in the previous character contrastive loss, which mistakenly causes the sparsity in the intra-class distribution and exacerbates ambiguity on challenging samples. Therefore, a new Character Unidirectional Alignment Loss is proposed to correct this error and unify the representation of the same characters in all samples by aligning the character features in the student model with the reference features in the teacher model. Extensive experiment results show that our method achieves SOTA performance (94.7\% and 70.9\% average accuracy on common benchmarks and Union14M-Benchmark). Code will be available at https://github.com/qqqyd/ViSu.