DRT: A Unified Framework for Text Detection, Recognition and Tracking in Video

The successfull analysis of video data is currently in great demand because a video is a major source of data in our lives. The text is a direct source information, while recent surveys on the detection of and recognition in imagery mainly focuses on extracting text scene pictures. Here, this work presents an implementation of text detection, video tracking and recognition with three Contributions. First, a unified generic framework is proposed for video text extraction that consistently implementing detection, recognition and tracking. Secondly, this framework is implemented recognize caption text and scene text. . This work proposes a new generic framework that successively implements text detection, recognition and tracking in videos, the text can be of any type, both caption text and scene text.