Adaptive classification and grading model of cigar wrapper leaf based on improved ResNet algorithm
Chaofan Du, Ruiqi Wang, Tianyi Wu, Yang Song, Runying Lin, Fulong Shen, Rijun Lai, Xiaolu Lin, Xiaofang Xie, Nannan Liu · 2026
Read Adaptive classification and grading model of cigar wrapper leaf based on improved ResNet algorithm by Chaofan Du, Ruiqi Wang, Tianyi Wu, Yang Song, Runying Lin, Fulong Shen, Rijun Lai, Xiaolu Lin, Xiaofang Xie, Nannan Liu (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-13478456.txt.
Description
Du and colleagues built an image-based grading system in 2026 for cigar wrapper leaves, the outer leaves that decide a cigar's look and much of its value. Grading still relies on hand sorting, which is slow and inconsistent between workers. Their dataset held 8,637 leaf photographs across nine standard purchasing grades for the FX-01 cultivar grown in Longyan, Fujian, and their model added a leaf-mask channel to a ResNet-50 image network.
On held-out test images the model reached 94.39 percent accuracy in assigning the correct grade to each leaf photographed. Its macro F1 score, balancing accuracy across grades, was 0.950, and its agreement-weighted kappa score reached 0.964 against human grades. Two training heads, one for plain classification and one respecting grade order, were fused with a 0.7 to 0.3 weighting, and the authors see a path to phone-based grading tools for buyers.
Catalog record
| CREATOR | Chaofan Du, Ruiqi Wang, Tianyi Wu, Yang Song, Runying Lin, Fulong Shen, Rijun Lai, Xiaolu Lin, Xiaofang Xie, Nannan Liu |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 36 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |