Seminar – Energy-Efficient Video Systems Design: Dedicated HW, Approximate Memories, and Near-Memory Processing

7 March 2023 – 1:30 PM
Politecnico di Torino – Maxwell Room

Please, remember that the event is free, but registration is required to attend.

The robust popularization of videos noticed along the last decades, allied to the omnipresence of smart mobile devices with multimedia-capable features, has led to intense development and research focusing on efficient video coding techniques, display technologies, and mobile device architectures. In this scenario, VVC (Versatile Video Coding) and AV1 (AOMedia Video 1) are key enablers of the future video systems by leading to meaningful data reduction through advanced coding techniques. However, real-time encoding/decoding for high-definition videos demands high processing effort, high memory communication and, consequently, extremely high energy consumption. As a result, the realization of real-time encoding/decoding targeting mobile systems has been posing serious challenges to industry and academia. This talk intends to briefly introduce the main concepts of video coding and present energy-efficient solutions to enable high-definition video encoding/decoding on mobile battery-powered embedded systems. In addition to the dedicated digital HW design perspective, we discuss the need for dedicated memory hierarchies featuring low-energy techniques such as near-memory processing and approximate memories.

Prof. Bruno Zatt

Bruno Zatt is a Professor at the Federal University of Pelotas (UFPel), Pelotas, Brazil, and a member of the Group of Architectures and Integrated Circuits (GACI) and the Video Technology Research Group (ViTech). He has 18 years research experience on algorithms and hardware architectures for video processing including 3 years as an intern researcher at the Karlsruhe Institute of Technology (KIT), Karlsruhe, Germany, and 1-year experience as a visiting professor at the University of California, Irvine, USA. Currently he is a visiting professor/ researcher at UniTo (Università di Torino) and Polito (Politecnido di Torino), Turin, Italy. He has published over 170+ papers in international journals/conferences and one book entitled “3D video coding for embedded devices”. His expertise includes visual signal processing and coding standards such as 2D video coding (H.264, HEVC, AV1, VVC), multiview video coding (MVC), 3D video coding (3D-HEVC) and Light Fields coding (JPEG Pleno). He holds the status of IEEE Senior Member (since 2016) and CNPq (National Council for Scientific and Technological Development) productivity research fellow (since 2017). He has coordinated several research projects funded by public (FAPERGS, CNPq, CAPES) and private funding agencies (Instituto Serrapilheira). Prof. Zatt also contributes as a member of the Brazilian Committee on Audio, Image, Multimedia and Hypermedia Coding and serves as IEEE T-CSVT (Transaction on Circuits and Systems for Video Technology) Associated Editor (2020-2023), IEEE OJ-SP (Open Journal of Signal Processing) Associated Editor, JICS (Journal of Integrated Circuits) Coeditor, and member of IEEE CASS Visual Signal Processing and Communication Technical Committee (VSPC-TC, 2020-2023).

Register here

Download the flyer: