Deep Learning

This paper proposes DRPCA-Net, a dynamic deep unfolding network for infrared small target detection, integrating sparsity-aware priors and dynamic parameter generation for robust performance.

This paper proposes SeqCSIST, a new task and benchmark for sequential closely-spaced infrared small target unmixing, and introduces DeRefNet for adaptive inter-frame information aggregation.

This paper introduces EV-UAV, the first large-scale event-based small object detection dataset for anti-UAV tasks, and proposes EV-SpSegNet as a novel baseline.

This paper proposes DeepPro, an efficient deep temporal probe network for IRST detection, leveraging global temporal profile information for superior performance.

This paper proposes DISTA-Net, the first deep learning model for unmixing closely-spaced infrared small targets, and introduces an open-source ecosystem for this field.

This paper proposes AuxDet, a multi-modal IRSTD framework that leverages auxiliary metadata for robust omni-domain infrared small target detection.

This paper proposes MoCoLSK, a modality-conditioned large selective kernel network for high-resolution LST downscaling, and introduces the GrokLST open-source ecosystem.