Dynamic scene understanding: The role of orientation features in space and time in scene classification

DMCA

Dynamic scene understanding: The role of orientation features in space and time in scene classification (2012)

Cached

Download Links

by Konstantinos G. Derpanis , Matthieu Lecce , Kostas Daniilidis , Richard P. Wildes

Venue:	In CVPR
Citations:	14 - 2 self

BibTeX

@INPROCEEDINGS{Derpanis12dynamicscene,
    author = {Konstantinos G. Derpanis and Matthieu Lecce and Kostas Daniilidis and Richard P. Wildes},
    title = {Dynamic scene understanding: The role of orientation features in space and time in scene classification},
    booktitle = {In CVPR},
    year = {2012}
}

OpenURL

Abstract

Natural scene classification is a fundamental chal-lenge in computer vision. By far, the majority of stud-ies have limited their scope to scenes from single image stills and thereby ignore potentially informative tempo-ral cues. The current paper is concerned with determin-ing the degree of performance gain in considering short videos for recognizing natural scenes. Towards this end, the impact of multiscale orientation measurements on scene classification is systematically investigated, as re-lated to: (i) spatial appearance, (ii) temporal dynamics and (iii) joint spatial appearance and dynamics. These measurements in visual space, x-y, and spacetime, x-y-t, are recovered by a bank of spatiotemporal oriented energy filters. In addition, a new data set is introduced that contains 420 image sequences spanning fourteen scene categories, with temporal scene information due to objects and surfaces decoupled from camera-induced ones. This data set is used to evaluate classification performance of the various orientation-related repre-sentations, as well as state-of-the-art alternatives. It is shown that a notable performance increase is realized by spatiotemporal approaches in comparison to purely spatial or purely temporal methods. 1.

Keyphrases

scene classification orientation feature dynamic scene understanding temporal scene information classification performance visual space multiscale orientation measurement informative tempo-ral cue fundamental chal-lenge joint spatial appearance spatial appearance energy filter new data set computer vision camera-induced one notable performance increase natural scene single image still short video natural scene classification performance gain temporal method current paper spatiotemporal approach various orientation-related repre-sentations fourteen scene category data set image sequence temporal dynamic state-of-the-art alternative

DMCA