Multispectral Deep Neural Networks for Pedestrian Detection

TL;DR

This paper investigates multispectral pedestrian detection using deep neural networks, demonstrating that fusing color and thermal images at middle convolutional layers significantly improves detection performance.

Contribution

It models multispectral pedestrian detection as a ConvNet fusion problem and designs four fusion architectures, with the Halfway Fusion model achieving the best results.

Findings

Halfway Fusion outperforms baseline by 11% in accuracy.

Fusion of color and thermal images enhances detection performance.

Middle-level feature fusion yields the best results among proposed architectures.

Abstract

Multispectral pedestrian detection is essential for around-the-clock applications, e.g., surveillance and autonomous driving. We deeply analyze Faster R-CNN for multispectral pedestrian detection task and then model it into a convolutional network (ConvNet) fusion problem. Further, we discover that ConvNet-based pedestrian detectors trained by color or thermal images separately provide complementary information in discriminating human instances. Thus there is a large potential to improve pedestrian detection by using color and thermal images in DNNs simultaneously. We carefully design four ConvNet fusion architectures that integrate two-branch ConvNets on different DNNs stages, all of which yield better performance compared with the baseline detector. Our experimental results on KAIST pedestrian benchmark show that the Halfway Fusion model that performs fusion on the middle-level convolutional features outperforms the baseline method by 11% and yields a missing rate 3.5% lower than the other proposed architectures.