IR-WSANet: A lightweight network with frequency-spatial joint optimization for infrared small target detection
Zhan Jin, Haonan Chen, Jiaxiang Chen · 2026
Read IR-WSANet: A lightweight network with frequency-spatial joint optimization for infrared small target detection by Zhan Jin, Haonan Chen, Jiaxiang Chen (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13473457.txt.
Description
Jin and co-authors present IR-WSANet, a lightweight neural network that spots tiny infrared targets, such as distant drones, against noisy backgrounds. Their 2026 paper rebuilds the YOLOv10 detector with wavelet-based convolution that separates fine detail from broad scene structure, plus a single-head attention module that weighs local and global cues together.
Detection accuracy jumped on all three test sets: mean average precision reached 97.2% on the SIDD-City set, 82.6% on SIDD-Mountain, and 82.8% on the HIT-UAV aerial set, beating the YOLOv10 baseline by 2.8 to 14.1% across the benchmarks. The single-head design also cut the attention parameter count by 27% against multi-head attention, keeping the model light enough for field hardware. The authors target early-warning and surveillance uses.
Catalog record
| CREATOR | Zhan Jin, Haonan Chen, Jiaxiang Chen |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 95 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |