SVTRv2: CTC Beats Encoder-Decoder Models in Scene Text Recognition

TL;DR

SVTRv2 is a scene text recognition model that outperforms encoder-decoder models by effectively handling text irregularities and leveraging linguistic context, while maintaining fast inference speed.

Contribution

The paper introduces SVTRv2, a CTC-based scene text recognition model with novel modules for text irregularity handling and linguistic context integration, improving accuracy without increasing inference time.

Findings

SVTRv2 surpasses most encoder-decoder models in accuracy.

SVTRv2 maintains fast inference speed.

The model performs well across various text irregularities and languages.

Abstract

Connectionist temporal classification (CTC)-based scene text recognition (STR) methods, e.g., SVTR, are widely employed in OCR applications, mainly due to their simple architecture, which only contains a visual model and a CTC-aligned linear classifier, and therefore fast inference. However, they generally exhibit worse accuracy than encoder-decoder-based methods (EDTRs) due to struggling with text irregularity and linguistic missing. To address these challenges, we propose SVTRv2, a CTC model endowed with the ability to handle text irregularities and model linguistic context. First, a multi-size resizing strategy is proposed to resize text instances to appropriate predefined sizes, effectively avoiding severe text distortion. Meanwhile, we introduce a feature rearrangement module to ensure that visual features accommodate the requirement of CTC, thus alleviating the alignment puzzle. Second, we propose a semantic guidance module. It integrates linguistic context into the visual features, allowing CTC model to leverage language information for accuracy improvement. This module can be omitted at the inference stage and would not increase the time cost. We extensively evaluate SVTRv2 in both standard and recent challenging benchmarks, where SVTRv2 is fairly compared to popular STR models across multiple scenarios, including different types of text irregularity, languages, long text, and whether employing pretraining. SVTRv2 surpasses most EDTRs across the scenarios in terms of accuracy and inference speed. Code: https://github.com/Topdu/OpenOCR.