Integrating prior knowledge about local contrast characteristics of infrared small targets into deep learning models can significantly enhance detection performance, as demonstrated by the novel LCAE-Net architecture.
Background semantics are crucial for accurately detecting small, clustered targets in infrared images, and the proposed BAFE-Net, trained on the novel DenseSIRST dataset, leverages this contextual information to significantly improve detection accuracy and reduce false alarms.
Combining self-supervised learning and a contrario reasoning within a YOLO object detection framework significantly improves infrared small target detection, particularly in challenging conditions with limited data.
This paper introduces a novel label evolution framework called LESPS (Label Evolution with Single Point Supervision) to address the challenge of efficient and accurate infrared small target detection using only point-level supervision, significantly reducing annotation costs while achieving comparable performance to fully supervised methods.
SeRankDet, a novel deep learning architecture, achieves superior infrared small target detection by employing selective rank-aware attention and dynamic feature fusion to overcome the limitations of traditional methods and enhance target-background separation.
The proposed Gradient Attention Network (GaNet) effectively extracts and preserves edge and gradient information of small infrared targets, while a global feature extraction module provides comprehensive background perception to improve detection performance.
The proposed Spatial-channel Cross Transformer Network (SCTransNet) leverages spatial-channel cross transformer blocks to effectively model the semantic differences between infrared small targets and complex backgrounds, enabling accurate detection of small targets.
A novel Scale and Location Sensitive (SLS) loss is proposed to improve infrared small target detection by addressing the limitations of existing loss functions in capturing scale and location information. A simple Multi-Scale Head is introduced to the plain U-Net (MSHNet) to leverage the SLS loss, achieving state-of-the-art performance without complex model structures.
Proposing a diffusion model framework for IRSTD to address target-level insensitivity and designing a low-frequency isolation module in the wavelet domain.
The author proposes a diffusion model framework for Infrared Small Target Detection to address target-level insensitivity by generating mask posterior distributions.