BEGIN:VCALENDAR
VERSION:2.0
PRODID:icalendar-ruby
CALSCALE:GREGORIAN
X-WR-CALNAME:Segmenting and tracking visual objects: a new computational th
 eory
X-WR-TIMEZONE:Eastern Time (US & Canada)
BEGIN:VEVENT
DTSTAMP:20260817T044955Z
UID:tag:localist.com\,2008:EventInstance_39005253163371
DTSTART:20220203T200000Z
DTEND:20220203T210000Z
DESCRIPTION:The survival of any animal depends on its ability to successful
 ly interact with the objects in its environment\, whether foraging for foo
 d\, building a shelter\, or identifying a conspecific. Thus\, persistent o
 bjects\, independently movable chunks of matter\, constitute the fundament
 al unit of organization in an animal’s ecological universe. How does the
  brain come to represent such chunks as distinct\, persistent wholes?\n\nI
  will discuss a new mathematical theory of persistent surface representati
 on (cf. https://arxiv.org/abs/2107.02036v1). The mathematical structure of
  light rays reflected from environment surfaces yields a natural represent
 ation of persistent surfaces\, and this surface representation provides a 
 solution to both the segmentation and tracking problems. I will describe h
 ow to generate this surface representation from continuous visual input\, 
 and present computational results showing how the approach can segment and
  invariantly track objects in the cluttered synthetic video despite severe
  appearance changes due to changes in viewpoint\, occlusion\, and object d
 eformation\, without requiring learning. Finally\, I will speculate on whe
 re and how this surface representation might be implemented by the primate
  visual system.
LOCATION:
SUMMARY:Segmenting and tracking visual objects: a new computational theory
URL;VALUE=URI:https://events.seas.harvard.edu/event/segmenting_and_tracking
 _visual_objects_a_new_computational_theory
CATEGORIES:Colloquia / Seminar / Lecture
END:VEVENT
END:VCALENDAR