Calendar of Events

Subscribe to our events calendar by RSS Feed Icon or iCal/Google Calendar iCal icon feeds.
(For Google Calendar, right-click on the icon, choose 'Copy link address', then visit your Google Calendar page and 'Add by URL.')

  • Ilan Kadar, Ben-Gurion University Download Ilan Kadar, Ben-Gurion University to my desktop calendar

    May 17, 2013 12:00 PM - 1:00 PM Ilan Kadar, Ben-Gurion University "From Perceptual Relations to Scene Gist Recognition" Abstract: The ability to recognize visual scenes quickly and accurately is highly constructive for both biological and machine vision. Following the seminal demonstrations of the ability of humans to recognize scenes in a fraction of a second, much research has been devoted to understanding its underlying visual process, as well as its computational modeling. In this talk we focus on one aspect of the scene gist recognition process and investigate whether prior knowledge of the perceptual relations between the different categories may help facilitate better scene understanding representation and operations. We first describe a psychophysical paradigm that probes human scene gist recognition and extracts perceptual relations between scene categories. Then, we introduce "SceneNet" - a new and comprehensive ontology database of scene categories derived directly from a large-scale human study that organizes scene categories according to their perceptual relations. This ontology database suggests that perceptual relations do not always conform to the semantic structure between categories, and provides a lower dimensional perceptual space with "perceptually meaningful" Euclidean distance, with each embedded scene category being represented by a single prototype. Next, we incorporate the SceneNet ontology into a computational scheme for learning non-linear mapping of scene images into the perceptual space, where each scene image is closer to its category prototype than to any other prototype by a large margin. In addition to much better computational results on various large scale scene understanding operations, the SceneNet ontology and perceptual space provides important insights into human scene representation and organization, and may serve as a key element in better understanding of this important perceptual capacity. Metcalf, room 305 Dept: CLPS, Departments