AI and the World's Hard Problems · Life sciences · Nature (computational biology) · 2026-10-26
Hybrid local–global vision transformer framework for robust plant disease detection in precision agriculture
Identifying plant diseases accurately matters for food security, but CNNs have limited receptive fields and vision transformers often lose fine local texture. The authors introduce ViT-LGFNet, which combines a local CNN stem, a transformer encoder and a cross-attention fusion module. On the PlantVillage dataset it reached a mean validation accuracy of 95.15% ± 0.74 across runs, with a best case of 96.28%.