BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Video-based Behavior Analysis
DTSTART:20180604T110000
DTEND:20180604T130000
DTSTAMP:20260407T183852Z
UID:39b7844c52b2edb5d2c811841ee067a2054ff93a8e3f4a08b552b8e9
CATEGORIES:Conferences - Seminars
DESCRIPTION:Stefano Savaré\nEDIC candidacy exam\nExam president: Prof. Ma
 rtin Jaggi\nThesis advisor: Prof. Pascal Fua\nCo-examiner: Dr. François F
 leuret\n\nAbstract\nModeling and analyzing people behavior in crowded scen
 es can greatly improve\ncomputer algorithms for scene understanding.\nIn t
 his report we review 3 different approaches used in this field.\nThe first
  is based on an advanced pedestrians motion model that keeps into account\
 nhuman-human\, human-crowds and human-scene interactions.\nThe second expl
 oits Long Short-Term Memory networks (LSTM) to learn the causalities behin
 d human\nmotion in a completely data-driven approach.\nThe third\, based o
 n the Lagrangian framework for fluid dynamics\, performs crowds analysis\n
 through a streakline representation  of the flow.\nFinally\, we suggest a
  new line of research to perform behavior analysis for groups of people.\n
 \nBackground papers\nPedestrian Behavior Modeling From Stationary Crowds W
 ith Applications to Intelligent Surveillance\, by Yi\, S.\, et al.\nLearni
 ng to predict human behaviour in crowded scenes\, by Alahi A.\, et al.\nA 
 Streakline Representation of Flow in Crowded Scenes\, by Mehran R.\, et al
 .\n 
LOCATION:BC 329 https://plan.epfl.ch/?room==BC%20329
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR