BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Video Super-Resolution with Convolutional Neural Networks
DTSTART:20161124T140000
DTEND:20161124T160000
DTSTAMP:20260502T005350Z
UID:ec9d4033c0b257f1ae004b6cdcc8840cd939af58bd16bc0ee1e7b939
CATEGORIES:Conferences - Seminars
DESCRIPTION:Ruofan Zhou\nEDIC Candidacy Exam\nExam President: Prof. Pascal
  Fua\nThesis Director: Prof. Sabine Süsstrunk\nCo-examiner: Prof. Wenzel 
 Jakob\n\nBackground papers\nA Bayesian Approach to Adaptive Video Super Re
 solution\,by Liu C.\, Sun\, D.\nSuper-resolution: a comprehensive survey\,
 by Nasrollahi K.\, Moeslund\, T.\nVideo Super-Resolution with Convolutiona
 l Neural Networks\, by Kappeler\, A.\, et al.\n\nAbstract  \nSuper-resolv
 ing real-world video sequences remains to be a challenging research task i
 n the field of computer vision. Many of the early works focus on model-bas
 ed methods such as maximum a posteriori and adaptive filtering that can ac
 hieve promising results but highly rely on the assumption on the data. As 
 nowadays data-driven method convolutional neural networks (CNN) have so fa
 r been successfully applied to image super-resolution (SR) as well as othe
 r restoration tasks\, we propose a CNN solution for video super-resolution
 . In this proposal\, we firstly present a conventional adaptive video supe
 r-resolution algorithm which uses a Bayesian approach [1]. Secondly\, we e
 xplore the early works of super-resolution [2] research field comprehensiv
 ely and discuss several key factors such as motion estimation\, data stati
 stics and evaluation metrics. Finally\, we review a recent work [3] that i
 ntroduce CNN in part of their video super-resolution framework. Based on t
 he ideas from these three papers and the results from our current work\, w
 e conclude with a discussion of potential further directions.\n 
LOCATION:BC 329 https://plan.epfl.ch/?room==BC%20329
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR