Authors: yft

Affiliation: henan university

Email: yftao@henu.edu.cn

Description: We propose a natural scene text detection algorithm based on multi-scale features and multi-objective functions. The scene image is fed into a Fully Convolutional Network (FCN) model and the output includes the coordinates of text line and two corresponding confidence scores. The bounding box of text which can be either rotated rectangles or quadrangles are sent to Non-Maximum Suppression to yield final results.