Novel Feature Fusion for Infrared Small Target Detection Feature Pyramid Networks

Abstract

Detecting infrared small target that lack texture features and shape information in cluttered environments is a challenging task. In this paper, we propose a novel feature fusion approach of feature pyramid networks (FPN) for effective detection of infrared small target. To extract the feature maps of infrared small target in different layers of the network and to fuse them effectively, we propose a multi-scale feature fusion module. Experimental results show that our proposed method performs much better than traditional approaches for infrared small target detection. In particular, our proposed method still achieves satisfactory results compared to other deep learning-based methods. In addition, we conducted ablation study of the network structure and the experimental results demonstrate the effectiveness of our proposed novel FPN.