Authors: opconty

Email: gaolijun8080@163.com

Description: we propose a new scene text detection method based on semantic features(A Semantic-based Arbitrarily-Oriented Scene Text Detector(named STD++ as it is the improved version of STD).
By reconstructing the text label, the coordinates of the center point of the small text box, the square length of the box and the center line of the text box are predicted directly on the feature map extracted from the Fully Convolutional Network(FCN).The proposed method does not require any complex pre-processing and pre-defined anchors, and the post-processing stage does not need to use non-maximum suppression. Only the Connected Component Labeling(CCL) is performed. Due to lack of computing resources and time.we only trained and tested with a single image resolution.