Improving End-to-End Text Image Translation From the Auxiliary Text Translation Task

TL;DR

This paper introduces a multi-task learning approach for text image translation that leverages auxiliary text translation and recognition tasks to improve performance, effectively utilizing large-scale text corpora.

Contribution

The novel integration of text translation as an auxiliary task in end-to-end text image translation enhances translation accuracy by sharing parameters and exploiting related tasks.

Findings

Outperforms existing end-to-end methods in experiments

Joint multi-task learning improves translation and recognition results

Auxiliary tasks are complementary and beneficial

Abstract

End-to-end text image translation (TIT), which aims at translating the source language embedded in images to the target language, has attracted intensive attention in recent research. However, data sparsity limits the performance of end-to-end text image translation. Multi-task learning is a non-trivial way to alleviate this problem via exploring knowledge from complementary related tasks. In this paper, we propose a novel text translation enhanced text image translation, which trains the end-to-end model with text translation as an auxiliary task. By sharing model parameters and multi-task training, our model is able to take full advantage of easily-available large-scale text parallel corpus. Extensive experimental results show our proposed method outperforms existing end-to-end methods, and the joint multi-task learning with both text translation and recognition tasks achieves better results, proving translation and recognition auxiliary tasks are complementary.