CORE-Text: Improving Scene Text Detection with Contrastive Relational Reasoning

TL;DR

CORE-Text introduces a contrastive relational reasoning module that enhances scene text detection by effectively modeling relations among text proposals, significantly reducing the sub-text problem.

Contribution

The paper proposes the CORE module, a novel relational reasoning approach that improves scene text detection by learning instance-aware representations through contrastive sub-text discrimination.

Findings

CORE-Text outperforms existing methods on four benchmarks.

The CORE module effectively mitigates the sub-text problem.

Enhanced relational reasoning improves detection accuracy.

Abstract

Localizing text instances in natural scenes is regarded as a fundamental challenge in computer vision. Nevertheless, owing to the extremely varied aspect ratios and scales of text instances in real scenes, most conventional text detectors suffer from the sub-text problem that only localizes the fragments of text instance (i.e., sub-texts). In this work, we quantitatively analyze the sub-text problem and present a simple yet effective design, COntrastive RElation (CORE) module, to mitigate that issue. CORE first leverages a vanilla relation block to model the relations among all text proposals (sub-texts of multiple text instances) and further enhances relational reasoning via instance-level sub-text discrimination in a contrastive manner. Such way naturally learns instance-aware representations of text proposals and thus facilitates scene text detection. We integrate the CORE module into a two-stage text detector of Mask R-CNN and devise our text detector CORE-Text. Extensive experiments on four benchmarks demonstrate the superiority of CORE-Text. Code is available: \url{https://github.com/jylins/CORE-Text}.