TY - CONF

T1 - The fastest pedestrian detector in the west

AU - Dollár, Piotr

AU - Belongie, Serge

AU - Perona, Pietro

PY - 2010

Y1 - 2010

N2 - We demonstrate a multiscale pedestrian detector operating in near real time (∼6 fps on 640×480 images) with state-of-the-art detection performance. The computational bottleneck of many modern detectors is the construction of an image pyramid, typically sampled at 8-16 scales per octave, and associated feature computations at each scale. We propose a technique to avoid constructing such a finely sampled image pyramid without sacrificing performance: our key insight is that for a broad family of features, including gradient histograms, the feature responses computed at a single scale can be used to approximate feature responses at nearby scales. The approximation is accurate within an entire scale octave. This allows us to decouple the sampling of the image pyramid from the sampling of detection scales. Overall, our approximation yields a speedup of 10-100 times over competing methods with only a minor loss in detection accuracy of about 1-2% on the Caltech Pedestrian dataset across a wide range of evaluation settings. The results are confirmed on three additional datasets (INRIA, ETH, and TUD-Brussels) where our method always scores within a few percent of the state-of-the-art while being 1-2 orders of magnitude faster. The approach is general and should be widely applicable.

AB - We demonstrate a multiscale pedestrian detector operating in near real time (∼6 fps on 640×480 images) with state-of-the-art detection performance. The computational bottleneck of many modern detectors is the construction of an image pyramid, typically sampled at 8-16 scales per octave, and associated feature computations at each scale. We propose a technique to avoid constructing such a finely sampled image pyramid without sacrificing performance: our key insight is that for a broad family of features, including gradient histograms, the feature responses computed at a single scale can be used to approximate feature responses at nearby scales. The approximation is accurate within an entire scale octave. This allows us to decouple the sampling of the image pyramid from the sampling of detection scales. Overall, our approximation yields a speedup of 10-100 times over competing methods with only a minor loss in detection accuracy of about 1-2% on the Caltech Pedestrian dataset across a wide range of evaluation settings. The results are confirmed on three additional datasets (INRIA, ETH, and TUD-Brussels) where our method always scores within a few percent of the state-of-the-art while being 1-2 orders of magnitude faster. The approach is general and should be widely applicable.

UR - http://www.scopus.com/inward/record.url?scp=84898465753&partnerID=8YFLogxK

U2 - 10.5244/C.24.68

DO - 10.5244/C.24.68

M3 - Paper

AN - SCOPUS:84898465753

T2 - 2010 21st British Machine Vision Conference, BMVC 2010

Y2 - 31 August 2010 through 3 September 2010

ER -