Speaker: Academician A. Murat Tekalp
Title: Recent Advances in Neural Video Compression
Time: 16:00 pm, August 16, 2022 (Tuesday)
Abstract:
Neural image video compression (also known as learned image/image compression) have recently exceeded the performance of traditional standards-based image video codecs. This is mainly because they allow end-to-end rate-distortion (R-D) optimized training of nonlinear transform, motion compensation and entropy models simultaneously. A further benefit of data-driven deep learning approach is that neural models can be optimized for any differentiable loss function, including visual perceptual loss functions, leading to perceptual image video compression, which cannot be easily handled by traditional codecs. This talk reviews the fundamentals of and recent advances in learned image video compression, including advances in motion-compensation models, flow residual coding, conditional coding, spatio-temporal entropy modeling, multi-rate neural models and the rate-perception-distortion tradeoff in learned image video coding. I will discuss the state-of-the-art in learned video compression and present recent results on learned low-delay and random access codec configurations, including our own work on hierarchical bi-directional video compression that combines the benefits of hierarchical bi-directional motion compensation and end-to-end rate-distortion optimization.
Personal Introduction:
A. Murat Tekalp (Fellow, IEEE) received the Ph.D. degree in electrical, computer, and systems engineering from Rensselaer Polytechnic Institute (RPI), Troy, NY, USA, in 1984. He was with Eastman Kodak Company, Rochester, New York, from 1984 to 1987, and with the University of Rochester, Rochester, New York, from 1987 to 2005, where he was promoted to Distinguished University Professor. He is currently a Professor with Koc University, Istanbul, Turkey. He served as the Dean of Engineering between 2010 and 2013. His research interests are in digital image and video processing, including video compression and streaming, video networking, multi-view and 3-D video processing, and deep learning for image video processing and compression. He has been elected a member of Turkish Academy of Sciences and Academia Europaea. He served as an Associate Editor for the IEEE Transactions on Signal Processing (1990–1992) and IEEE Transactions on Image Processing (1994–1996). He was the Editor-in-Chief for the EURASIP journal Signal Processing: Image Communication published by Elsevier between 1999 and 2010. He was on the Editorial Board for the IEEE Signal Processing Magazine (2007–2010) and the Proceedings of the IEEE (2014–2020). He chaired the IEEE Signal Processing Society Technical Committee on Image and Multidimensional Signal Processing (January 1996–December 1997). He was appointed as the General Chair of IEEE International Conference on Image Processing (ICIP), Rochester, NY, USA, in 2002 and the Technical Program Co-Chair for IEEE ICIP 2020 and ICIP 2024. He served in the European Research Council (ERC) Advanced Grant Panels (2009–2015) and Starting Grant Panel in 2021. He is currently on the Editorial Board of Wiley-IEEE Press. He has authored the Prentice Hall book Digital Video Processing (1995), a completely rewritten second edition of which is published in 2015.
[Editor: Xiaohan Liu]