BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Saliency In Comics
DTSTART:20190828T100000
DTEND:20190828T120000
DTSTAMP:20260406T230136Z
UID:2156ca11c68f7585198a6d12ce7790eb51b9ae3e5245629e175017f6
CATEGORIES:Conferences - Seminars
DESCRIPTION:Bahar Aydemir\nEDIC candidacy exam\nExam president: Prof. Pier
 re Dillenbourg\nThesis advisor: Prof. Sabine Süsstrunk\nCo-examiner: Dr. 
 Mathieu Salzmann\n\nAbstract\nHuman attention in visual contexts is a wide
 ly researched topic which gives clues about the execution process of the
  human visual system and cognition. Although deep learning methods improv
 ed the performances of saliency detection models\, the performances appea
 r to be saturated in the last years. Bylinskii et al. explore the commonl
 y missed regions by these models and suggest using high-level concepts to
  describe them [1]. Moreover\, the models that benefit from deep learning
  methods require large scale datasets. Jiang et al. describe a paradigm 
 to collect large scale saliency data on natural images [2]. This paradigm 
 mimics the human foveal vision by representing the content in different r
 esolutions. Even though salient regions in natural settings are studied b
 y many\, there are few approaches for the detection of salient regions in
  comics images. Thirunarayanan et al\, present a method to find and segme
 nt important parts on legacy comics by using eye-tracking and generated 
 gaze points [3]. Lastly\, we introduce our research plan to detect salien
 t regions on comics by using crowdsourcing and eye-tracking. \n\n\nBackg
 round papers\nWhere Should Saliency Models Look Next?\, by Bylinskii Z.\,
  et al.\nSALICON: Saliency in Context \, by  Jiang M.\, et al.\nCreatin
 g Segments and Effects on Comics by Clustering Gaze Data \, by Thirunaray
 anan I.\, et al.\n\n 
LOCATION:BC 329 https://plan.epfl.ch/?room==BC%20329
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR