Events Calendar

Premkumar Natarajan: Some Recent Advances in Offline Handwriting Recognition

Speaker: Premkumar Natarajan, USC Information Sciences Institute

Talk Title: Some Recent Advances in Offline Handwriting Recognition

Abstract: For the past three decades, the task of automatically transcribing the text content of handwritten hardcopy documents (usually referred to as offline handwriting recognition) has remained a thorny challenge. Until a few years ago, research efforts in offline recognition have focused exclusively on script-dependent recognition approaches that are developed for a specific target language or script. Progress was spotty and recognition accuracies poor on real-world data. Starting in 2007, with sponsorship from the DARPA MADCAT program, we started developing a script-independent methodology for offline handwriting recognition which has since yielded revolutionary improvements in offline handwriting recognition performance. The research results have had significant impact at the recent NIST OpenHART 2013 (Open Handwriting Evaluation) workshop, all the participants employed our script-independent methodology in their submissions. In this talk, I will present some recent advances we have accomplished in handwriting recognition and also provide a (very) brief overview of salient historical context. In the latter part of the talk, I will discuss current trends and some open research tasks that might be of interest to some in the audience. I will end the presentation with a discursive discussion of some salient research directions at ISI.

Biography: Premkumar (Prem) Natarajan is the Executive Director of the Information Sciences Institute (ISI) at the University of Southern California and a Vice Dean of Engineering in the Viterbi School of Engineering. In his current role, Prem sets the technical vision and operational strategy for ISI. Previously, he served as an Executive Vice President and Principal Scientist at Raytheon BBN Technologies where he oversaw technical and business operations in Speech, Language and Multimedia Technologies at BBN.

Prem�s technical contributions span a wide range of multimedia processing and pattern recognition areas, including optical character recognition (OCR, speech recognition, speech-to-speech translation, video analysis and content extraction, topic classification, and speech triage). He has served as Principal Investigator or Senior Advisor on numerous Department of Defense (DOD) and Intelligence Community sponsored research and deployment projects, including important DARPA-sponsored research efforts such as the TRANSTAC, MADCAT, and DEFT programs; IARPA-sponsored research efforts such as VACE and ALADDIN; and other significant USG-sponsored efforts such as the Army MFLTS Program of Record. He is an active member of the professional communities in document and speech processing, and serves on academic advisory boards.

Host: Gaurav Sukhatme

Location: Henry Salvatori Computer Science Center (SAL) - 101

Audiences: Everyone Is Invited

Contact: Assistant to CS chair