Infrared Small Target Detection with Scale and Location Sensitivity

TL;DR

This paper introduces a novel Scale and Location Sensitive (SLS) loss for infrared small target detection, improving localization and scale sensitivity, and proposes a simple Multi-Scale Head network that outperforms existing methods.

Contribution

The paper presents a new SLS loss that enhances IRSTD by focusing on scale and location sensitivity, combined with a simple MSHNet architecture for improved detection performance.

Findings

SLS loss improves detection accuracy over traditional losses.

MSHNet outperforms state-of-the-art methods on IRSTD tasks.

SLS loss enhances generalization of existing detectors.

Abstract

Recently, infrared small target detection (IRSTD) has been dominated by deep-learning-based methods. However, these methods mainly focus on the design of complex model structures to extract discriminative features, leaving the loss functions for IRSTD under-explored. For example, the widely used Intersection over Union (IoU) and Dice losses lack sensitivity to the scales and locations of targets, limiting the detection performance of detectors. In this paper, we focus on boosting detection performance with a more effective loss but a simpler model structure. Specifically, we first propose a novel Scale and Location Sensitive (SLS) loss to handle the limitations of existing losses: 1) for scale sensitivity, we compute a weight for the IoU loss based on target scales to help the detector distinguish targets with different scales: 2) for location sensitivity, we introduce a penalty term based on the center points of targets to help the detector localize targets more precisely. Then, we design a simple Multi-Scale Head to the plain U-Net (MSHNet). By applying SLS loss to each scale of the predictions, our MSHNet outperforms existing state-of-the-art methods by a large margin. In addition, the detection performance of existing detectors can be further improved when trained with our SLS loss, demonstrating the effectiveness and generalization of our SLS loss. The code is available at https://github.com/ying-fu/MSHNet.