Learning to Predict More Accurate Text Instances for Scene Text Detection

TL;DR

This paper introduces a pixel-based scene text detection method that improves arbitrary shape text instance prediction by using a novel coordinate regression loss and an accuracy refinement loss, achieving state-of-the-art results.

Contribution

The paper proposes a new pixel-based detection framework with a target vertex sorting-free regression loss and an accuracy loss guided by IoU for better arbitrary shape text detection.

Findings

Achieves 84.8% F-measure on Total-Text benchmark.

Outperforms existing methods on arbitrary shape and multi-oriented text detection.

Demonstrates effectiveness of the proposed losses in improving detection accuracy.

Abstract

At present, multi-oriented text detection methods based on deep neural network have achieved promising performances on various benchmarks. Nevertheless, there are still some difficulties for arbitrary shape text detection, especially for a simple and proper representation of arbitrary shape text instances. In this paper, a pixel-based text detector is proposed to facilitate the representation and prediction of text instances with arbitrary shapes in a simple manner. Firstly, to alleviate the effect of the target vertex sorting and achieve the direct regression of arbitrary shape text instances, the starting-point-independent coordinates regression loss is proposed. Furthermore, to predict more accurate text instances, the text instance accuracy loss is proposed as an assistant task to refine the predicted coordinates under the guidance of IoU. To evaluate the effectiveness of our detector, extensive experiments have been carried on public benchmarks which contain arbitrary shape text instances and multi-oriented text instances. We obtain 84.8% of F-measure on Total-Text benchmark. The results show that our method can reach state-of-the-art performance.