Implementation of a Cascaded Histogram of Oriented Gradient (HOG)-Based Pedestrian Detector

Report No. ARL-TR-6661
Authors: Christopher Reale, Prudhvi Gurram, Shuowen Hu, and Alex Chan
Date/Pages: September 2013; 36 pages
Abstract: In this report, we present our implementation of a cascaded Histogram of Oriented Gradient (HOG) based pedestrian detector. Most human detection algorithms can be implemented as a cascade of classifiers to decrease computation time while maintaining approximately the same performance. Although cascaded versions of Dalal and Triggss HOG detector already exist, we aim to provide a more detailed explanation of an implementation than is currently available. We also use Asymmetric Boosting instead of Adaboost to train the cascade stages. We show that this approach reduces the number of weak classifiers needed per stage. We present the results of our detector on the INRIA pedestrian detection dataset and compare them to the results provided by Dalal and Triggs.
Distribution: Approved for public release
  Download Report ( 0.929 MBytes )
If you are visually impaired or need a physical copy of this report, please visit and contact DTIC.

Last Update / Reviewed: September 1, 2013