Abstract:Segmenting small infrared targets presents a significant challenge for traditional image processing architectures due to the inherent lack of texture, minimal shape information, and their sparse pixel representation within images. The conventional UNet architecture, while proficient in general segmentation tasks, inadequately addresses the nuances of small infrared target segmentation due to its reliance on downsampling operations, such as pooling, which often results in the loss of critical target information. This paper introduces the Channel Spatial Attention Nested UNet (CSAN-UNet), an innovative architecture designed specifically to enhance the detection and segmentation of small infrared targets. Central to CSAN-UNet's design is the Cascaded Channel and Spatial Convolutional Attention Module (CSCAM), a novel component that adaptively enhances multi-level features and mitigates the loss of target information attributable to downsampling processes. Additionally, the Channel-priority and Spatial Attention Cascade Module (CPSAM) represents another pivotal advancement within CSAN-UNet, prioritizing channel-level adjustments alongside spatial attention mechanisms to efficiently extract deep semantic information pertinent to small infrared targets. Empirical validation conducted on two public datasets confirms that CSAN-UNet surpasses existing state-of-the-art algorithms in segmentation performance, while simultaneously reducing computational overhead.

What problem does this paper attempt to address?

The paper primarily addresses the challenges in infrared small target detection, particularly the limitations of traditional image processing architectures when dealing with such targets, such as lack of texture, insufficient shape information, and sparse pixel representation. The authors propose a new network structure—Channel and Spatial Attention Nested UNet (CSAN-UNet), aimed at improving the detection and segmentation performance of small infrared targets. Specifically, the study addresses the following key issues: 1. **Improving Detection Accuracy**: Traditional UNet architectures have deficiencies when handling small infrared targets because they rely on downsampling operations (e.g., pooling), which often lead to the loss of important target information. Therefore, the paper proposes CSAN-UNet, which enhances multi-level features and mitigates the loss of target information due to downsampling by introducing a Cascade Channel and Spatial Convolutional Attention Module (CSCAM). 2. **Optimizing Feature Extraction**: To efficiently extract key semantic information related to small infrared targets, the paper also develops a lightweight Channel-Priority and Spatial Attention Cascade Module (CPSAM). This module can effectively map spatial relationships at a low computational cost, ensuring precise extraction of target information. 3. **Balancing Performance and Efficiency**: By combining the advantages of CSCAM and CPSAM, CSAN-UNet can effectively retain detailed target features while maintaining computational resource efficiency. This addresses the issue in existing methods where either the computational demand is too high or small object features may be lost. 4. **Empirical Validation**: The authors demonstrate through empirical validation on two public datasets that CSAN-UNet has significant advantages in segmentation performance compared to existing state-of-the-art algorithms, while also reducing computational overhead. In summary, the main contribution of this study is the proposal of an innovative network structure aimed at improving the accuracy of infrared small target detection while considering computational efficiency. Through careful design of the network structure, especially the introduction of CSCAM and CPSAM, CSAN-UNet has made significant progress in addressing the complex problem of infrared small target detection.

CSAN-UNet: Channel Spatial Attention Nested UNet for Infrared Small Target Detection

An Improved U-Net Infrared Small Target Detection Algorithm Based on Multi-Scale Feature Decomposition and Fusion and Attention Mechanism

Dense Nested Attention Network for Infrared Small Target Detection

A Multi-Task Framework for Infrared Small Target Detection and Segmentation

CCRANet: A Two-Stage Local Attention Network for Single-Frame Low-Resolution Infrared Small Target Detection

Infrared Small Target Detection Based on Local Contrast Vector and Signed Normalization

iSmallNet: Densely Nested Network with Label Decoupling for Infrared Small Target Detection

SCINet: Spatial and Contrast Interactive Super-Resolution Assisted Infrared UAV Target Detection

Location-Guided Dense Nested Attention Network for Infrared Small Target Detection

UCDnet: Double U-Shaped Segmentation Network Cascade Centroid Map Prediction for Infrared Weak Small Target Detection

Hierarchical attention-guided multiscale aggregation network for infrared small target detection

Improved Dense Nested Attention Network Based on Transformer for Infrared Small Target Detection

SFA-UNet: More Attention to Multi-Scale Contrast and Contextual Information in Infrared Small Object Segmentation

MCAT-UNet: Convolutional and Cross-Shaped Window Attention Enhanced UNet for Efficient High-Resolution Remote Sensing Image Segmentation

An Infrared Small Target Detection Method Using Segmentation Based Region Proposal and CNN

Infrared Small Target Detection Based on a Temporally-Aware Fully Convolutional Neural Network

Aware Distribute and Sparse Network for Infrared Small Target Detection

An infrared small target detection method using coordinate attention and feature fusion

Lost in UNet: Improving Infrared Small Target Detection by Underappreciated Local Features

Unbalanced Class Learning Network With Scale-Adaptive Perception for Complicated Scene in Remote Sensing Images Segmentation

ISPANet: A Pyramid Self-Attention Network for Single-Frame High-Resolution Infrared Small Target Detection With a Large-Scale Dataset SHR-IRST