Scene Text Recognition with Temporal Convolutional Encoder

TL;DR

This paper introduces a Temporal Convolutional Encoder for scene text recognition that captures long-term dependencies, improving accuracy over traditional sequence-to-sequence models.

Contribution

The paper proposes a novel Temporal Convolutional Encoder that enhances scene text recognition by modeling long-term temporal dependencies within the encoder stage.

Findings

Improved recognition accuracy on seven datasets.

Effectiveness of attention modules in convolutional blocks.

Temporal Convolutional Encoder outperforms existing methods.

Abstract

Texts from scene images typically consist of several characters and exhibit a characteristic sequence structure. Existing methods capture the structure with the sequence-to-sequence models by an encoder to have the visual representations and then a decoder to translate the features into the label sequence. In this paper, we study text recognition framework by considering the long-term temporal dependencies in the encoder stage. We demonstrate that the proposed Temporal Convolutional Encoder with increased sequential extents improves the accuracy of text recognition. We also study the impact of different attention modules in convolutional blocks for learning accurate text representations. We conduct comparisons on seven datasets and the experiments demonstrate the effectiveness of our proposed approach.