ILNet: Low-Level Matters for Salient Infrared Small Target Detection

Infrared small target detection is a technique for finding small targets from infrared clutter background. Previous deep-learning-based approaches have achieved promising results. However, the lack of high-level semantic information leads to a degradation of small infrared target features in the deeper layers of the neural network, resulting in suboptimal representation capabilities. To address this issue, we propose an Infrared Low-level Network (ILNet) that conceptualizes infrared small targets as salient regions characterized by limited semantic information. In contrast to other state-of-the-art methods, ILNet emphasizes low-level information more significantly, rather than treating it uniformly with high-level information. A lightweight feature fusion module, named Interactive Polarized Orthogonal Fusion module (IPOF), is proposed, which integrates more important low-level features from the shallow layers into the deep layers. A Dynamic One-Dimensional Aggregation layers (DODA) are inserted into the IPOF, to dynamically adjust the aggregation of low dimensional information according to the number of input channels. In addition, the idea of ensemble learning is used to design a Representative Block (RB) to dynamically allocate weights for shallow and deep layers. Experimental results on the challenging NUAA-SIRST (78.22% nIoU and $1.33 \times 10^{-6}$ Fa) and IRSTD-1k (68.91% nIoU and $3.23 \times 10^{-6}$Fa) dataset demonstrate that the proposed ILNet can get better performances than other SOTA methods. Moreover, ILNet can obtain a greater improvement with the increasement of data volume. Training code are available at https://github.com/Li-Haoqing/ILNet.