R2CNN: Rotational Region CNN for Orientation Robust Scene Text Detection

TL;DR

This paper introduces R2CNN, a novel rotation-aware scene text detection method based on Faster R-CNN, capable of accurately detecting arbitrarily oriented texts in natural images, achieving competitive benchmark results.

Contribution

The paper presents R2CNN, a new rotation-invariant text detection framework that extends Faster R-CNN with inclined bounding box prediction and inclined non-maximum suppression.

Findings

Achieves competitive results on ICDAR 2015 and ICDAR 2013 benchmarks.

Effectively detects arbitrarily oriented texts in natural scene images.

Outperforms previous methods in orientation robustness.

Abstract

In this paper, we propose a novel method called Rotational Region CNN (R2CNN) for detecting arbitrary-oriented texts in natural scene images. The framework is based on Faster R-CNN [1] architecture. First, we use the Region Proposal Network (RPN) to generate axis-aligned bounding boxes that enclose the texts with different orientations. Second, for each axis-aligned text box proposed by RPN, we extract its pooled features with different pooled sizes and the concatenated features are used to simultaneously predict the text/non-text score, axis-aligned box and inclined minimum area box. At last, we use an inclined non-maximum suppression to get the detection results. Our approach achieves competitive results on text detection benchmarks: ICDAR 2015 and ICDAR 2013.