Partially occluded weed classification using vision transformers and convolutional neural networks for precision agriculture
Euan Hall, Emmanuel Junior Zuza, Karen Rial Lovera, Chris McCarthy · 2026
Read Partially occluded weed classification using vision transformers and convolutional neural networks for precision agriculture by Euan Hall, Emmanuel Junior Zuza, Karen Rial Lovera, Chris McCarthy (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/PMC13575115.txt.
Description
Euan Hall and three colleagues compare image models for spotting weeds that are partly hidden by crops, in a 2026 study for precision farming. Automated weed detection lets farmers spray only where needed, cutting chemical use. Hidden leaves remain a weak point for field systems.
The team tested two vision transformer models against two standard convolutional networks under set levels of hiding: none, 25 percent, and 50 percent. One transformer, ViT-B16, grew more accurate as hiding increased, reaching 86 percent at the highest level. The other models fell back, landing between 45 and 76 percent. Statistical checks put model design as the main driver of accuracy. The authors suggest hybrid designs that blend transformer reach with cheaper convolutional processing.
Catalog record
| CREATOR | Euan Hall, Emmanuel Junior Zuza, Karen Rial Lovera, Chris McCarthy |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 57276 |
| LICENSE | https://creativecommons.org/licenses/by/4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |