ARBITRARY-ORIENTED SCENE TEXT DETECTION VIA ROTATION PROPOSALS

Abstract

This paper introduces a novel rotation-based framework for arbitrary-oriented text detection in natural scene images. We present the Rotation Region Proposal Networks (RRPN), which are designed to generate inclined proposals with text orientation angle information. The angle information is then adapted for bounding box regression to make the proposals more accurately fit into the text region in terms of the orientation. The Rotation Region-of-Interest (RRoI) pooling layer is proposed to project arbitrary-oriented proposals to a feature map for a text region classifier. The whole framework is built upon a region proposal- based architecture, which ensures the computational efficiency of the arbitrary-oriented text detection compared with previous text detection systems. We conduct experiments using the rotation-based framework on three real-world scene text detection datasets and demonstrate its superiority in terms of effectiveness and efficiency over previous approaches.

Existing System

Fully convolutional network (FCN)

Proposed System

In this paper, we develop a rotation-based approach and an end-to-end text detection system for arbitrary-oriented text detection. Particularly, orientations are incorporated so that the detection system can generate proposals for arbitrary orientation. A comparison between the previous horizontal based approach and ours is illustrated. We present the Rotation Region Proposal Networks (RRPN), which are designed to generate inclined proposals with text orientation angle information. The angle information is then adapted for bounding box regression to make the proposals more accurately fit the text region. The Rotation Region-of-Interest (RRoI) pooling layer is proposed to project arbitrary-oriented proposals to a feature map. Finally, a two-layer network is deployed to classify the regions as either text or background.

Architecture

BLOCK DIAGRAM