Harbin Institute's DEDM-Net Achieves Breakthrough in Infrared Small Target Detection

Researchers from the Research Center for Space Optical Engineering at Harbin Institute of Technology have unveiled a groundbreaking deep learning network designed to detect tiny, low-contrast infrared targets that often elude existing methods. Their work, published (DOI: 10.34133/remotesensing.1046) on June 30, 2026, in the Journal of Remote Sensing, introduces the Diffusion-Enhanced Dense Mamba Network (DEDM-Net), a two-stage architecture that synergistically combines diffusion-based feature enhancement with state-space modeling to suppress background clutter and amplify target signals. The challenge is formidable: infrared small targets typically occupy fewer than 81 pixels, exhibit signal-to-noise ratios around 3, and lack distinct shape or texture, making them easily submerged in complex backgrounds. Traditional deep learning methods often focus solely on target features, neglecting the vast background and exacerbating class imbalance. DEDM-Net tackles this by modeling both targets and backgrounds simultaneously.

The first stage employs a dual-path diffusion model with a novel blind processing module that predicts each pixel using only surrounding information—never the pixel itself—preventing extremely small targets from being misclassified as background. The second stage introduces a dense nested Mamba architecture based on the state space model (SSM), which captures long-range correlations across global and local features with linear computational complexity, a significant advantage over conventional Transformers. A cross-stage prediction fusion module further integrates features from both stages, improving contour segmentation accuracy. Evaluated on three public datasets—NUAA-SIRST, NUDT-SIRST, and IRSTD-1k—DEDM-Net achieved state-of-the-art performance. On NUDT-SIRST, it reached 93.40% IoU, 93.28% nIoU, 98.37% detection probability, and a false-alarm rate of just 3.75×10⁻⁶, outperforming DNA-Net and ISTDU-Net. On IRSTD-1k, it achieved 73.71% IoU and 93.89% detection probability with only 11.10×10⁻⁶ false alarms, surpassing all competitors.

Corresponding author Dr. Shikai Jiang explained, “Infrared small targets are extremely challenging because they lack shape and texture—they're essentially just a few bright pixels in a sea of background. By modeling both the target-free background and potential target regions simultaneously, our diffusion-enhanced approach effectively amplifies what matters while suppressing what doesn't. The Mamba architecture then provides the global context needed to distinguish true targets from bright clutter.” While DEDM-Net achieves superior accuracy, the diffusion-based two-stage design increases inference time compared to single-stage networks. Future work will focus on model distillation, mixed-precision inference, and faster samplers to reduce the required diffusion steps. The approach holds promise for real-time surveillance systems, autonomous drone navigation in low-visibility conditions, and early wildfire detection networks. As the team noted, the framework could also inspire new thinking about how generative models and state-space architectures can be combined for other challenging computer vision tasks where target-background separation is critical.