FH-Net: Frequency-domain and hierarchical dual enhancement for few-shot surface defect segmentation
Lili Fan, Zhanglin Yang, Fan Guo, Ping Fang · 2026
Read FH-Net: Frequency-domain and hierarchical dual enhancement for few-shot surface defect segmentation by Lili Fan, Zhanglin Yang, Fan Guo, Ping Fang (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/plos-pone-0358329.txt.
Description
Lili Fan and three colleagues built a vision system for factory inspection, published in PLOS ONE in 2026. Spotting surface defects usually needs many labeled examples, but new products and rare flaws give inspectors only a handful. Their network, called FH-Net, is designed to learn from just a few samples.
Fan's method works in the frequency domain and refines features in layers, a dual boost that helps it find defect regions with little training. Few-shot learning like this suits real production lines, where stopping to gather thousands of images is not an option. The study speaks to quality control across manufacturing, from electronics to textiles, wherever defects are rare but costly.
Catalog record
| CREATOR | Lili Fan, Zhanglin Yang, Fan Guo, Ping Fang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 77791 |
| LICENSE | CC_BY |
| RIGHTS | CC BY |
| SOURCE | PLOS |