ARFC-WAHNet: Adaptive Receptive Field Convolution and Wavelet-Attentive Hierarchical Network for Infrared Small Target Detection

TL;DR

The paper introduces ARFC-WAHNet for infrared small target detection, integrating adaptive receptive field convolution (MRFFIConv) with dynamic multi-branch processing, wavelet-based downsampling (WFED), and cross-level attention modules (HLFF, GMEA). The approach addresses ISTD challenges—tiny target size, low contrast, and clutter—by preserving fine details and enhancing target features across scales. Empirical results on SIRST, NUDT-SIRST, and IRSTD-1K show state-of-the-art performance with favorable efficiency, and ablations confirm the individual and synergistic contributions of MRFFIConv, WFED, HLFF, and GMEA. The work advances ISTD by combining dynamic convolutions, frequency-domain enhancement, and global attention to improve robustness in complex scenes, with publicly available code for reproducibility.

Abstract

Infrared small target detection (ISTD) is critical in both civilian and military applications. However, the limited texture and structural information in infrared images makes accurate detection particularly challenging. Although recent deep learning-based methods have improved performance, their use of conventional convolution kernels limits adaptability to complex scenes and diverse targets. Moreover, pooling operations often cause feature loss and insufficient exploitation of image information. To address these issues, we propose an adaptive receptive field convolution and wavelet-attentive hierarchical network for infrared small target detection (ARFC-WAHNet). This network incorporates a multi-receptive field feature interaction convolution (MRFFIConv) module to adaptively extract discriminative features by integrating multiple convolutional branches with a gated unit. A wavelet frequency enhancement downsampling (WFED) module leverages Haar wavelet transform and frequency-domain reconstruction to enhance target features and suppress background noise. Additionally, we introduce a high-low feature fusion (HLFF) module for integrating low-level details with high-level semantics, and a global median enhancement attention (GMEA) module to improve feature diversity and expressiveness via global attention. Experiments on public datasets SIRST, NUDT-SIRST, and IRSTD-1k demonstrate that ARFC-WAHNet outperforms recent state-of-the-art methods in both detection accuracy and robustness, particularly under complex backgrounds. The code is available at https://github.com/Leaf2001/ARFC-WAHNet.

Figures (13)

• Figure 1: Comparison of IoU, parameter count (#Params.), and FLOPs of mainstream ISTD deep learning methods on the NUDT-SIRST dataset r35.
• Figure 2: The overall architecture of the proposed ARFC-WAHNet, which has a U-Net framework with MRFFConv, WFED, HLFF, and GMEA modules.
• Figure 3: The structure of horizontal and vertical central difference convolution (HV-CDC) and diagonal central difference convolution (DG-CDC).
• Figure 4: Illustration of the MRFFIConv module.
• Figure 5: Illustration of a heatmap. The columns from left to right represent the original image, ground truth, heatmap output from Regular Conv, MSDC, MDDC, and DCN, respectively.