Handwriting Recognition – Still Some Way to go for Computers
Brown Bag Lecture by Dr. Szilárd Vajda | 5/29/2012 11AM-12PM | 7th Floor Conference Room, Bldg 38A
Abstract: Is automatic handwriting recognition a reality or still hype? The first part of the talk will discuss the importance of recognizing handwriting in historical documents, forms, postal documents, whiteboards and tablets. We highlight the scientific challenges from image processing to labeling, spotting and recognition. The second part is a brief review of research results in Bangla (Bengali) postal document automation, whiteboard reading using an adaptive layout model, some recent work done in historical documents, and in the recognition of a little known Indonesian script, never addressed before in the literature. Finally, some issues related to machine learning will be discussed such as feature learning and unbalanced data.
Bio: Dr. Vajda, joined CEB in early April 2012, and concentrates his research in pattern recognition, machine learning with special application to document analysis, handwriting recognition and related topics such as classification, supervised/unsupervised/semi-supervised learning, clustering methods, automatic feature learning strategies and human-computer interaction. Previously, he worked at Robotics Research Institute, Technical University of Dortmund (Germany), Furukawa Electric Institute of Technology, Budapest, Hungary, and Loria Research Center, Nancy, France. Dr. Vajda holds a B.S. in Computer Science from Babeș-Bolyai University, Cluj-Napoca, Romania and a Ph.D. in Computer Science from Henri Poincaré University, Nancy, France.