SEED: Semantics Enhanced Encoder-Decoder Framework for Scene Text Recognition

TL;DR

This paper introduces a semantics enhanced encoder-decoder framework for scene text recognition that incorporates global semantic information to improve robustness against low-quality images, outperforming existing methods.

Contribution

It integrates semantic information into encoder-decoder models, exemplified by ASTER, to enhance recognition accuracy on challenging scene texts.

Findings

Outperforms state-of-the-art on benchmark datasets

More robust to image blur and illumination issues

Effective integration of semantic info improves recognition

Abstract

Scene text recognition is a hot research topic in computer vision. Recently, many recognition methods based on the encoder-decoder framework have been proposed, and they can handle scene texts of perspective distortion and curve shape. Nevertheless, they still face lots of challenges like image blur, uneven illumination, and incomplete characters. We argue that most encoder-decoder methods are based on local visual features without explicit global semantic information. In this work, we propose a semantics enhanced encoder-decoder framework to robustly recognize low-quality scene texts. The semantic information is used both in the encoder module for supervision and in the decoder module for initializing. In particular, the state-of-the art ASTER method is integrated into the proposed framework as an exemplar. Extensive experiments demonstrate that the proposed framework is more robust for low-quality text images, and achieves state-of-the-art results on several benchmark datasets.