Boosting up Scene Text Detectors with Guided CNN

TL;DR

This paper introduces Guided CNN, a framework that improves scene text detection accuracy and speed by using a guidance mask to focus computation on text regions, demonstrated with state-of-the-art methods.

Contribution

The paper proposes Guided CNN, a novel framework that enhances text detection efficiency and accuracy by integrating a guidance subnetwork with a primary detector, and introduces a new training strategy.

Findings

Speeds up CTPN by 2.9 times on ICDAR 2013 with 1.5% F-measure improvement.

Speeds up EAST by 2.0 times on ICDAR 2015 with 1.0% F-measure improvement.

Effective and efficient framework demonstrated with two state-of-the-art methods.

Abstract

Deep CNNs have achieved great success in text detection. Most of existing methods attempt to improve accuracy with sophisticated network design, while paying less attention on speed. In this paper, we propose a general framework for text detection called Guided CNN to achieve the two goals simultaneously. The proposed model consists of one guidance subnetwork, where a guidance mask is learned from the input image itself, and one primary text detector, where every convolution and non-linear operation are conducted only in the guidance mask. On the one hand, the guidance subnetwork filters out non-text regions coarsely, greatly reduces the computation complexity. On the other hand, the primary text detector focuses on distinguishing between text and hard non-text regions and regressing text bounding boxes, achieves a better detection accuracy. A training strategy, called background-aware block-wise random synthesis, is proposed to further boost up the performance. We demonstrate that the proposed Guided CNN is not only effective but also efficient with two state-of-the-art methods, CTPN and EAST, as backbones. On the challenging benchmark ICDAR 2013, it speeds up CTPN by 2.9 times on average, while improving the F-measure by 1.5%. On ICDAR 2015, it speeds up EAST by 2.0 times while improving the F-measure by 1.0%.