Section: New Results

Pedestrian Recognition using Convolutional Neural Networks

Participants : Danut-Ovidiu Pop, Fawzi Nashashibi.

Pedestrian detection is of highly importance for a large number of applications, especially in the elds of automotive safety, robotics and surveillance. In spite of the widely varying methods developed in recent years, pedestrian detection is still an open challenge whose accuracy and robustness has to be improved. This year we focused on the improvement of the classification component in the pedestrian detection task by adopting two approaches: 1) by combining three image modalities (intensity, depth and ow) to feed a unique convolutional neural network (CNN) and 2) by fusing the results of three independent CNNs. The evaluations have been performed on the Daimler stereo vision data set.