Author(s)

Yang Dongming, Ge Shuiying

Corresponding Author

Yang Dongming

Abstract

The pedestrian detection methods based on convolutional networks showed a more advantage than methods with artificial features. However, most algorithms do not achieve the desired results because they have not made full use of convolutional features. In this paper, we proposed a model called Multiple Fused CNNs, which utilizes fusional feature maps from multi-scale convolutional layers to detect pedestrians.

Keywords

Pedestrian detection，Convolutional network，Multiscale，Fusional features.