A multi-scale small object detection framework for tomato leaf pest and disease detection in complex natural environments
Rui Gao, Shuang Ma, Meng Wang, Junyong Zhang, Jie Yang, Zhuoran Zhang, Shuai Yan, Dongrui Han · 2026
Read A multi-scale small object detection framework for tomato leaf pest and disease detection in complex natural environments by Rui Gao, Shuang Ma, Meng Wang, Junyong Zhang, Jie Yang, Zhuoran Zhang, Shuai Yan, Dongrui Han (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-13613504.txt.
Description
Spotting tiny pests and early disease spots on tomato leaves, against cluttered field backgrounds, defeats many vision models. The authors propose PLD-YOLO, a detection framework tuned for exactly this problem. A high-resolution feature branch protects the small details that deep networks usually wash out. Multi-scale convolutions and an adaptive mechanism widen the model's view of context while damping background noise. A dynamic weighting scheme gives small objects more say during training, where they are usually drowned by larger ones.
The team tests the system on a tomato dataset gathered in natural settings, against two-stage detectors, other YOLO versions, and transformer-based models. It leads on precision, recall, and the standard accuracy measures while staying fast. Ablation tests, cross-dataset checks, and varied field conditions support the design choices. The authors present the framework as a practical tool for smart plant protection and precision farming.
Catalog record
| CREATOR | Rui Gao, Shuang Ma, Meng Wang, Junyong Zhang, Jie Yang, Zhuoran Zhang, Shuai Yan, Dongrui Han |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 56 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |