Prof. Dr. Koichi Kise

"Logging Reading Activities with the help of Character Recognition and Document Image Retrieval "

Reading is an important and primal source of information for people. Therefore, tapping the information flow from the document to the reader is fruitful to know more about the reader, e.g., to find the reader's interests and preferences, as well as to recognize knowledge acquired by the user. We have started the project "reading-life log" to realize this functionality.

We are approaching to the reading-life log based on the first person vision for documents. By using a mobile eyetracker and a scene camera mounted on the reader, we are able to capture what the reader is reading. Camera-based character recognition and document image retrieval allow us to obtain symbol information from the camera captured image of the read part.

During my talk I will show you demos of character recognition and document image retrieval both of which work in real-time.