In this paper we describe an approach to automati- cally improving the efficiency of soft cascade-based person detec- tors. Our technique addresses the two fundamental bottlenecks in cascade detectors: the number of weak classifiers that need to be evaluated in each cascade, and the total number of detection windows to be evaluated. By simply observing a soft cascade operating on a scene, we learn scale specific linear approximations of cascade traces that allows us to eliminate a large fraction of the classifier evaluation. Independently, this time by observing regions of support in the soft cascade on a training set, we learn a coarse geometric model of the scene that allows our detector to propose candidate detection windows and significantly reduce the number of windows run through the cascade. Our approaches are unsupervised and require no additional labeled person images for learning. Our linear cascade approximation results in about 28% savings in detection, while our geometric model gives a saving of over 95%, without appreciable loss of accuracy.
Unsupervised scene adaptation for faster multi-scale pedestrian detection / BARTOLI, FEDERICO; LISANTI, GIUSEPPE; KARAMAN, SVEBOR; BAGDANOV, ANDREW DAVID; DEL BIMBO, ALBERTO. - ELETTRONICO. - (2014), pp. 3534-3539. (Intervento presentato al convegno International Conference on Pattern Recognition tenutosi a Stockholm, Sweden nel 2014) [10.1109/ICPR.2014.608].
Unsupervised scene adaptation for faster multi-scale pedestrian detection
LISANTI, GIUSEPPE;
2014
Abstract
In this paper we describe an approach to automati- cally improving the efficiency of soft cascade-based person detec- tors. Our technique addresses the two fundamental bottlenecks in cascade detectors: the number of weak classifiers that need to be evaluated in each cascade, and the total number of detection windows to be evaluated. By simply observing a soft cascade operating on a scene, we learn scale specific linear approximations of cascade traces that allows us to eliminate a large fraction of the classifier evaluation. Independently, this time by observing regions of support in the soft cascade on a training set, we learn a coarse geometric model of the scene that allows our detector to propose candidate detection windows and significantly reduce the number of windows run through the cascade. Our approaches are unsupervised and require no additional labeled person images for learning. Our linear cascade approximation results in about 28% savings in detection, while our geometric model gives a saving of over 95%, without appreciable loss of accuracy.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.