Which and Where to Focus: A Simple yet Accurate Framework for Arbitrary-Shaped Nearby Text Detection in Scene Images

Scene text detection has drawn the close attention of researchers. Though many methods have been proposed for horizontal and oriented texts, previous may not perform well when dealing with arbitrary-shaped texts such as curved texts. In particular, confusion problem arises in case nearby instances. this paper, we propose a simple yet effective method accurate scene detection. Firstly, One-to-Many Training Scheme (OMTS) is designed to eliminate enable proposals learn more appropriate groundtruths Secondly, Proposal Feature Attention Module (PFAM) exploit features each proposal, which can better adapt Finally, baseline that based on Faster R-CNN outputs curve representation directly. Equipped PFAM OMTS, detector achieve state-of-the-art or competitive performance several challenging benchmarks.