Pedestrian detection using deep channel features in monocular image sequences

Abstract

In this paper, we propose the Deep Channel Features as an extension to Channel Features for pedestrian detection. Instead of using hand-crafted features, our method automatically learns deep channel features as a mid-level feature by using a convolutional neural network. The network is pretrained by the unsupervised sparse filtering and a group of filters is learned for each channel. Combining the learned deep channel features with other low-level channel features (i.e. LUV channels, gradient magnitude channel and histogram of gradient channels) as the final feature, a boosting classifier with depth-2 decision tree as the weak classifier is learned. Our method achieves a significant detection performance on public datasets (i.e. INRIA, ETH, TUD, and CalTech).