DeepLabV3+ with MobileNetV3 backbone enhancement and multi-level fusion decoder for pepper field weed segmentation
Jun Lu, Fan Wang, Donglin Cao, Tianshu Jin · 2026
Read DeepLabV3+ with MobileNetV3 backbone enhancement and multi-level fusion decoder for pepper field weed segmentation by Jun Lu, Fan Wang, Donglin Cao, Tianshu Jin (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/PMC13473395.txt.
Description
Lu and colleagues present a 2026 computer vision model that separates pepper plants from weeds pixel by pixel in field images, a step toward targeted weeding. The task is hard because crop and weed share green textures, overlap each other, and appear at many sizes under changing light.
Their design improves a known segmentation network with a lightweight backbone and added modules that sharpen feature selection and rebuild edges. On an independent pepper field test set, the model reached about 96 percent overlap accuracy and about 98 percent recall, meaning it missed very few weed pixels. It also scored high on boundary quality and ran at about 47 frames per second. The authors say the speed and accuracy suit practical precision weeding machines.
Catalog record
| CREATOR | Jun Lu, Fan Wang, Donglin Cao, Tianshu Jin |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 52 KB |
| LICENSE | https://creativecommons.org/licenses/by/4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |