Reproduce “Seeing Through the Noise: Improving Infrared Small Target Detection and Segmentation from Noise Suppression Perspective” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Maoxun Yuan, Duanni Meng, Ziteng Xi, Tianyi Zhao, Shiji Zhao, Yimian Dai, Xingxing Wei

Researchers from Beihang University and Nankai University developed the Noise-Suppression Feature Pyramid Network (NS-FPN) to enhance infrared small target detection and segmentation. The NS-FPN achieves state-of-the-art performance by actively suppressing noise during feature fusion, resulting in significantly lower false alarm rates while maintaining high detection probability and segmentation accuracy across benchmark datasets.

Key insights

  • High-frequency image components are crucial for target localization but are also highly susceptible to noise interference, directly contributing to increased false alarms.
  • Low-frequency components, despite lacking fine detail for localization, possess a superior ability to suppress noise and can guide the purification of high-frequency features.
  • The unique Gaussian-like intensity distribution of infrared small targets allows for a structured, spiral-aware sampling approach to effectively extract target-relevant features and mitigate background noise.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.