Chinese and English bilingual scene text detection

In this paper, we develop an end-to-end framework for text region proposal generation and text detection in the natural scene, which takes advantage of the efficient training strategies of the neural network. We make some technical changes over the SSD architectures, so that our method can handle the Chinese and English bilingual situation. Experiments show that our method exceeds the state-of-the-art performance in ICDAR and MSRA-TD500 datasets. Furthermore, we compare the influence of factors such as NMS in detail on the experimental results.