Lightweight deep learning for tomato disease detection: trends, challenges, and edge AI perspectives

Summary

This review synthesises recent advances in lightweight deep learning and edge AI for tomato disease detection, emphasising practical deployment in precision agriculture. Experimental validation using MobileNetV2 and EfficientNetB0 on prevalent Tamil Nadu tomato diseases achieved 99.9% accuracy. The authors propose an integrated framework combining AI-powered diagnosis with microbial biocontrol, offering a scalable, farmer-friendly approach to sustainable disease management.

Regional applicability

The study focuses on tomato diseases prevalent in Tamil Nadu, India, and may have limited direct applicability to United Kingdom tomato production systems, which face different pathogen pressures and climatic conditions. However, the lightweight deep learning methodologies and edge AI deployment strategies described are transferable to UK horticulture contexts, particularly for resource-constrained or remote monitoring scenarios.

Key measures

Overall accuracy (99.9%), macro-F1 score (0.99) for MobileNetV2 and EfficientNetB0 models on tomato disease classification

Outcomes reported

The study evaluated lightweight deep learning models (MobileNetV2 and EfficientNetB0) for detecting common tomato diseases prevalent in Tamil Nadu, achieving 99.9% accuracy and macro-F1 score of 0.99. A framework combining AI-powered disease diagnosis with microbial biocontrol recommendations was proposed for sustainable, region-specific disease management.

Topic tags