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

Haoqing Li,Jinfu Yang,Runshi Wang,Yifei Xu

2023-09-24

Abstract:Infrared small target detection is a technique for finding small targets from infrared clutter background. Due to the dearth of high-level semantic information, small infrared target features are weakened in the deep layers of the CNN, which underachieves the CNN's representation ability. To address the above problem, in this paper, we propose an infrared low-level network (ILNet) that considers infrared small targets as salient areas with little semantic information. Unlike other SOTA methods, ILNet pays greater attention to low-level information instead of treating them equally. A new 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 1.33e-6 Fa) and IRSTD-1K (68.91% nIoU and 3.23e-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 <a class="link-external link-https" href="https://github.com/Li-Haoqing/ILNet" rel="external noopener nofollow">this https URL</a>.

Computer Vision and Pattern Recognition

What problem does this paper attempt to address?

The paper primarily focuses on addressing the technical challenges in Infrared Small Target Detection (IRST), particularly the issues encountered when dealing with small infrared targets in complex cluttered backgrounds. Specifically, the paper investigates the following key issues:

1. **Loss of High-Resolution Information**: Due to the lack of sufficient high-level semantic information for infrared small targets in deep Convolutional Neural Networks (CNNs), these target features are weakened in deep networks, leading to underutilization of the CNN's performance capabilities.

2. **Insufficient Low-Level Features**: Infrared small targets are usually small in size (less than 15x15 pixels), and their low-level features (such as color and texture) are insufficient, while high-level semantic features are often difficult to identify.

3. **Inherent Limitations of Infrared Images**: Compared to visible light images, infrared images have a lower Signal-to-Clutter Ratio (SCR) and more noise, making small targets easily submerged in the background and difficult to identify.

To address the above challenges, the paper proposes a new method called ILNet (Infrared Low-level Network). The key contributions of ILNet include:

- **Redefining the Problem Perspective**: Treating infrared small target detection as a salient object detection problem, considering infrared small targets as high-intensity salient regions in a cluttered background without specific semantic information.

- **Interactive Polarized Orthogonal Fusion Module (IPOF)**: Designing a module for bidirectional feature fusion, improving feature fusion through high-resolution interaction in both channel and spatial dimensions.

- **Dynamic One-Dimensional Aggregation Layer (DODA)**: Proposing a layer that can adaptively aggregate features based on the dimensions of the input features, retaining important information and details to enhance detection performance.

- **Representative Block (RB)**: Proposing a module that distinguishes the importance of high-level and low-level information to improve the feature attenuation problem in deep networks and dynamically fuse global features.

Experimental results show that ILNet performs excellently on two challenging infrared small target detection datasets, NUAA-SIRST and IRSTD-1K, with significant improvements over existing state-of-the-art (SOTA) methods in metrics such as IoU, nIoU, detection probability (Pd), and false alarm rate (Fa). Additionally, ILNet demonstrates a good balance between computational efficiency and accuracy.