BEGIN:VCALENDAR VERSION:2.0 PRODID:IEEE vTools.Events//EN CALSCALE:GREGORIAN BEGIN:VTIMEZONE TZID:US/Eastern BEGIN:DAYLIGHT DTSTART:20230312T030000 TZOFFSETFROM:-0500 TZOFFSETTO:-0400 RRULE:FREQ=YEARLY;BYDAY=2SU;BYMONTH=3 TZNAME:EDT END:DAYLIGHT BEGIN:STANDARD DTSTART:20231105T010000 TZOFFSETFROM:-0400 TZOFFSETTO:-0500 RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=11 TZNAME:EST END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTAMP:20230505T144847Z UID:D308AE2C-058F-40FF-9179-92B31BD1A3E7 DTSTART;TZID=US/Eastern:20230503T120000 DTEND;TZID=US/Eastern:20230503T130000 DESCRIPTION:Video coding is a fundamental and ubiquitous technology in the modern society. Generations of international video coding standards\, such as the widely-deployed H.264/AVC and H.265/HEVC and the latest H.266/VVC\ , provide essential means for enabling video conferencing\, video streamin g\, video sharing\, e-commerce\, entertainment\, and many more video appli cations. These existing standards all rely on the fundamental theory of si gnal processing and information theory to encode generic video efficiently with a favorable rate distortion behavior.\n\nIn recent years\, rapid adv ancement in deep learning and artificial intelligence technology has allow ed people to manipulate images and videos using deep generative models. Am ong these\, of particular interest to the field of video coding is the app lication of deep generative models towards compressing talking-face video at ultra-low bit rates. By focusing on talking faces\, generative models c an effectively learn the inherent structure about composition\, movement a nd posture of human faces and deliver promising results using very little bandwidth resource. At ultra-low bit rates\, when even the latest video co ding standard H.266/VVC is apt to suffer from significant blocking artifac ts and blurriness beyond the point of recognition\, generative methods can maintain clear facial features and vivid expression in the reconstructed video. Further\, generative face video coding techniques are inherently ca pable of manipulating the reconstructed face and promise to deliver a more interactive experience.\n\nIn this talk\, we start with a quick overview of traditional and deep learning-based video coding techniques. We then fo cus on face video coding with generative networks\, and present two scheme s that send different deep information in the bitstream\, one sending comp act temporal motion features and the other sending 3D facial semantics. We compare their compression efficiency and visual quality with that of the latest H.266/VVC standard\, and showcase the power of deep generative mode ls in preserving vivid facial images with little bandwidth resource. We al so present visualization results to exhibit the capability of the 3D facia l semantics-based scheme in terms of interacting with the reconstructed fa ce video and animating virtual faces.\n\nCo-sponsored by: Fairleigh Dickin son University\n\nSpeaker(s): Dr. Yan Ye\, \n\nAgenda: \nVideo coding is a fundamental and ubiquitous technology in the modern society. Generations of international video coding standards\, such as the widely-deployed H.26 4/AVC and H.265/HEVC and the latest H.266/VVC\, provide essential means fo r enabling video conferencing\, video streaming\, video sharing\, e-commer ce\, entertainment\, and many more video applications. These existing stan dards all rely on the fundamental theory of signal processing and informat ion theory to encode generic video efficiently with a favorable rate disto rtion behavior.\n\nIn recent years\, rapid advancement in deep learning an d artificial intelligence technology has allowed people to manipulate imag es and videos using deep generative models. Among these\, of particular in terest to the field of video coding is the application of deep generative models towards compressing talking-face video at ultra-low bit rates. By f ocusing on talking faces\, generative models can effectively learn the inh erent structure about composition\, movement and posture of human faces an d deliver promising results using very little bandwidth resource. At ultra -low bit rates\, when even the latest video coding standard H.266/VVC is a pt to suffer from significant blocking artifacts and blurriness beyond the point of recognition\, generative methods can maintain clear facial featu res and vivid expression in the reconstructed video. Further\, generative face video coding techniques are inherently capable of manipulating the re constructed face and promise to deliver a more interactive experience.\n\n In this talk\, we start with a quick overview of traditional and deep lear ning-based video coding techniques. We then focus on face video coding wit h generative networks\, and present two schemes that send different deep i nformation in the bitstream\, one sending compact temporal motion features and the other sending 3D facial semantics. We compare their compression e fficiency and visual quality with that of the latest H.266/VVC standard\, and showcase the power of deep generative models in preserving vivid facia l images with little bandwidth resource. We also present visualization res ults to exhibit the capability of the 3D facial semantics-based scheme in terms of interacting with the reconstructed face video and animating virtu al faces.\n\nVirtual: https://events.vtools.ieee.org/m/352355 LOCATION:Virtual: https://events.vtools.ieee.org/m/352355 ORGANIZER:tan@fdu.edu SEQUENCE:2 SUMMARY:Face Video Compression with Generative Models URL;VALUE=URI:https://events.vtools.ieee.org/m/352355 X-ALT-DESC:Description: <br /><p>Video coding is a fundamental and ubiquito us technology in the modern society. Generations of international video co ding standards\, such as the widely-deployed H.264/AVC and H.265/HEVC and the latest H.266/VVC\, provide essential means for enabling video conferen cing\, video streaming\, video sharing\, e-commerce\, entertainment\, and many more video applications. These existing standards all rely on the fun damental theory of signal processing and information theory to encode gene ric video efficiently with a favorable rate distortion behavior.</p>\n<p>I n recent years\, rapid advancement in deep learning and artificial intelli gence technology has allowed people to manipulate images and videos using deep generative models. Among these\, of particular interest to the field of video coding is the application of deep generative models towards compr essing talking-face video at ultra-low bit rates. By focusing on talking f aces\, generative models can effectively learn the inherent structure abou t composition\, movement and posture of human faces and deliver promising results using very little bandwidth resource. At ultra-low bit rates\, whe n even the latest video coding standard H.266/VVC is apt to suffer from si gnificant blocking artifacts and blurriness beyond the point of recognitio n\, generative methods can maintain clear facial features and vivid expres sion in the reconstructed video.&nbsp\; Further\, generative face video co ding techniques are inherently capable of manipulating the reconstructed f ace and promise to deliver a more interactive experience.</p>\n<p>In this talk\, we start with a quick overview of traditional and deep learning-bas ed video coding techniques. We then focus on face video coding with genera tive networks\, and present two schemes that send different deep informati on in the bitstream\, one sending compact temporal motion features and the other sending 3D facial semantics. We compare their compression efficienc y and visual quality with that of the latest H.266/VVC standard\, and show case the power of deep generative models in preserving vivid facial images with little bandwidth resource. We also present visualization results to exhibit the capability of the 3D facial semantics-based scheme in terms of interacting with the reconstructed face video and animating virtual faces .&nbsp\;</p><br /><br />Agenda: <br /><p>Video coding is a fundamental and ubiquitous technology in the modern society. Generations of international video coding standards\, such as the widely-deployed H.264/AVC and H.265/ HEVC and the latest H.266/VVC\, provide essential means for enabling video conferencing\, video streaming\, video sharing\, e-commerce\, entertainme nt\, and many more video applications. These existing standards all rely o n the fundamental theory of signal processing and information theory to en code generic video efficiently with a favorable rate distortion behavior.< /p>\n<p>In recent years\, rapid advancement in deep learning and artificia l intelligence technology has allowed people to manipulate images and vide os using deep generative models. Among these\, of particular interest to t he field of video coding is the application of deep generative models towa rds compressing talking-face video at ultra-low bit rates. By focusing on talking faces\, generative models can effectively learn the inherent struc ture about composition\, movement and posture of human faces and deliver p romising results using very little bandwidth resource. At ultra-low bit ra tes\, when even the latest video coding standard H.266/VVC is apt to suffe r from significant blocking artifacts and blurriness beyond the point of r ecognition\, generative methods can maintain clear facial features and viv id expression in the reconstructed video.&nbsp\; Further\, generative face video coding techniques are inherently capable of manipulating the recons tructed face and promise to deliver a more interactive experience.</p>\n<p >In this talk\, we start with a quick overview of traditional and deep lea rning-based video coding techniques. We then focus on face video coding wi th generative networks\, and present two schemes that send different deep information in the bitstream\, one sending compact temporal motion feature s and the other sending 3D facial semantics. We compare their compression efficiency and visual quality with that of the latest H.266/VVC standard\, and showcase the power of deep generative models in preserving vivid faci al images with little bandwidth resource. We also present visualization re sults to exhibit the capability of the 3D facial semantics-based scheme in terms of interacting with the reconstructed face video and animating virt ual faces.&nbsp\;</p> END:VEVENT END:VCALENDAR