Selective Attention-Based Network for Robust Infrared Small Target Detection

2026-05-05 · Source: cs.CV updates on arXiv.org · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems · Depth: Expert, extended

Summary

SANet, a Selective Attention-based Network, significantly advances infrared small target detection (IRSTD) by addressing limitations in existing deep learning methods. Built on the U-Net framework, SANet introduces two novel components: a Dual-path Semantic-aware Module (DSM) and a Selective Attention Fusion Module (SAFM). The DSM integrates standard convolutions for local detail with pinwheel-shaped convolutions for expanded, direction-sensitive receptive fields, enhanced by a Convolutional Block Attention Module (CBAM). The SAFM replaces static skip connections with a spatially adaptive, learnable weighting mechanism for context-aware, cross-scale feature fusion. Evaluated on NUAA-SIRST, IRSTD-1K, and NUDT-SIRST benchmarks, SANet consistently outperforms fourteen state-of-the-art methods, achieving IoU improvements of 1.93%, 4.32%, and 2.21% over the second-best approaches, demonstrating strong generalization and practical applicability.

Key takeaway

For research scientists developing advanced IRSTD systems, SANet offers a robust solution to persistent challenges like low signal-to-clutter ratios and complex backgrounds. You should consider integrating its Dual-path Semantic-aware Module and Selective Attention Fusion Module to enhance fine-grained target perception and dynamically fuse multi-scale features, potentially leading to significant improvements in detection accuracy and false alarm suppression in your applications.

Key insights

SANet enhances infrared small target detection by dynamically fusing multi-scale features and refining early feature extraction.

Principles

• Expanded receptive fields improve dim target perception.
• Dynamic attention fusion surpasses static skip connections.
• Spatial-channel recalibration refines feature representations.

Method

SANet uses a Dual-path Semantic-aware Module (DSM) for feature extraction and a Selective Attention Fusion Module (SAFM) for adaptive cross-scale feature fusion within a U-Net architecture, optimized with Soft-IoU loss.

In practice

• Deploy SANet for improved maritime surveillance.
• Integrate SANet into early warning systems.
• Apply SANet in precision-guided strike systems.

Topics

• Infrared Small Target Detection
• Selective Attention Networks
• Dual-path Semantic-aware Module
• Selective Attention Fusion Module
• Pinwheel-shaped Convolutions

Best for: Research Scientist, AI Scientist, Machine Learning Engineer, Computer Vision Engineer

Related on AIssential

• Frequency-Enhanced Dual-Subspace Networks for Few-Shot Fine-Grained Image Classification — FEDSNet improves few-shot fine-grained image classification by integrating frequency domain features to enhance structural stability and reduce overfitting.
• ASGNet: Adaptive Spectrum Guidance Network for Automatic Polyp Segmentation — Integrating spectral features with global attributes enhances polyp segmentation accuracy beyond spatial-only methods.
• Physics-Guided Attention in a Lightweight TCN for Efficient WiFi CSI-Based Human Activity Recognition — Incorporating physics-guided inductive biases into lightweight TCNs significantly improves WiFi CSI-based HAR efficiency and performance.
• Learning with Semantic Priors: Stabilizing Point-Supervised Infrared Small Target Detection via Hierarchical Knowledge Distillation — Hierarchical knowledge distillation with VFM semantics stabilizes point-supervised infrared small target detection.
• RISE: Single Static Radar-based Indoor Scene Understanding — RISE utilizes multipath radar reflections and generative AI for privacy-preserving, geometry-aware indoor scene understanding from a single static sensor.
• NERVE: A Neuromorphic Vision and Radar Ensemble for Multi-Sensor Fusion Research — NERVE is a multi-sensor dataset bridging neuromorphic vision and radar for robust perception research.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.CV updates on arXiv.org.