Text spotting in the wild

Our Siyang Qin advanced to candidacy with a Ph.D. thesis proposal focusing on automatic text detection in images.

Congrats Siyang!

 Text spotting in the wild

Systems that can automatically detect the presence of text in an image (text spotters) may find application in multiple practical scenarios, such as video surveillance, forensic, video annotation, and mobile OCR. May main interest in text spotting stems from its potential application as an assistive device for blind people. In spite of tremendous prorgress over the past year, detecting text in challenging situation with poor quality image, small scale text and confounding background is still an open problem. Taking advantage of the power of deep neural network, I am building a robust and effcient text spotting system which not only achieves state-of-the-art result on standard benchmarks, but also performs well in real world scenario. In future work, I will add a text recognition module and further improve speed and performance for video input, in order to build a real-time scene text detection and recognition system.