Fast human detection with cascaded ensembles on the GPU

Author

Bilgic, Berkin ; Horn, Berthold K P ; Masaki, Ichiro

Author_Institution

Dept. of Electr. Eng. & Comput. Sci., MIT, Cambridge, MA, USA

fYear

2010

fDate

21-24 June 2010

Firstpage

325

Lastpage

332

Abstract

We investigate a fast pedestrian localization framework that integrates the cascade-of-rejectors approach with the Histograms of Oriented Gradients (HoG) features on a data parallel architecture. The salient features of humans are captured by HoG blocks of variable sizes and locations which are chosen by the AdaBoost algorithm from a large set of possible blocks. We use the integral image representation for histogram computation and a rejection cascade in a sliding-windows manner, both of which can be implemented in a data parallel fashion. Utilizing the NVIDIA CUDA framework to realize this method on a Graphics Processing Unit (GPU), we report a speed up by a factor of 13 over our CPU implementation. For a 1280×960 image our parallel technique attains a processing speed of 2.5 to 8 frames per second depending on the image scanning density, which is similar to the recent GPU implementation of the original HoG algorithm in.

Keywords

computer graphic equipment; coprocessors; feature extraction; image representation; object detection; parallel architectures; AdaBoost algorithm; GPU; NVIDIA CUDA framework; data parallel architecture; graphics processing unit; histograms of oriented gradient; human detection; image representation; image scanning density; pedestrian localization; sliding windows; Central Processing Unit; Concurrent computing; Detectors; Graphics; Histograms; Humans; Intelligent vehicles; Support vector machine classification; Support vector machines; USA Councils;

fLanguage

English

Publisher

ieee

Conference_Titel

Intelligent Vehicles Symposium (IV), 2010 IEEE

Conference_Location

San Diego, CA

ISSN

1931-0587

Print_ISBN

978-1-4244-7866-8

Type

conf

DOI

10.1109/IVS.2010.5548145

Filename

5548145