Authors: Hao Wu, Na He, Zhou Shen, Dan Meng, Qingfeng Wang
Description: We used an intuitive region-based method to detect multi-oriented text without any prior knowledge regarding the textual shape. The proposals generated by our network are geometry adaptive, which makes our method robust to various text aspect ratios and orientations. And for those over-sized pictures , we will do some preprocess for them , like Gaussian blur or image pyramid . At last, some tricks are used to filter the results.