TextContourNet: a Flexible and Effective Framework for Improving Scene Text Detection Architecture with a Multi-task Cascade
This work addresses scene text detection, a domain-specific task in computer vision, with incremental improvements through a novel framework.
The paper tackles the problem of improving scene text detection by extracting text instance contour information from images, and demonstrates that using this as a multi-task cascade achieves the best performance, with experimental results showing improvements over a state-of-the-art detector.
We study the problem of extracting text instance contour information from images and use it to assist scene text detection. We propose a novel and effective framework for this and experimentally demonstrate that: (1) A CNN that can be effectively used to extract instance-level text contour from natural images. (2) The extracted contour information can be used for better scene text detection. We propose two ways for learning the contour task together with the scene text detection: (1) as an auxiliary task and (2) as multi-task cascade. Extensive experiments with different benchmark datasets demonstrate that both designs improve the performance of a state-of-the-art scene text detector and that a multi-task cascade design achieves the best performance.