Optimized Deep Learning for Enhanced Tomato Plant Disease Detection and Crop Health Management with Intelligent Systems

Tomatoes are central to India's cuisine, but they confront a significant threat: emerging diseases. Recent agriculture data highlights the issue, showing a distressing 30% decrease in tomato yields over a decade, equating to a staggering annual loss of approximately 6 million metric tonnes and $2.5 billion in revenue. Tomatoes provide health benefits, crucial for combating vitamin deficiencies, especially in malnourished regions. While convolutional neural network (CNN) models hold promise in disease diagnosis, they demand costly computational resources, making them impractical for agriculture. Traditional methods struggle with disease detection, achieving low sensitivity and specificity rates (around 70–75%). This research develops a cost-effective illness detection system that combines deep learning techniques and a unique voting strategy. Our approach uses a deep learning-based CNN model with EfficientNet, DenseNet, and Inception-V3 architectures for disease classification, achieving impressive accuracy: EfficientNet at 98.34%, Inception-V3 at 96.99%, and DenseNet at 95.92%. This work advances early disease detection and promises to boost tomato crop yields significantly. Projections indicate a potential 20% increase in yields for resource-constrained farmers. Ultimately, this innovation aims to enhance food production while minimizing computational overhead, making a substantial contribution to the field.