Smart Agriculture: A Deep Learning Model for Plant Disease Diagnosis

Accurate early stage detection of crop diseases is essential for maintaining crop quality and yield by enabling timely and effective treatments. However, disease detection often requires specialized expertise in plant pathology and significant experience. Therefore, an automated crop disease detection system is invaluable in agriculture, as it supports the development of an early warning system. To build this system, we developed a stepwise disease detection model utilizing images of diseased and healthy plants, along with a CNN based algorithm incorporating five pre trained models. The detection process involves three stages: crop classification, disease detection, and disease classification. An ‘unknown’ category was included to enhance model generalization for broader applications. In validation tests, the model achieved high accuracy in classifying crops and disease types (97.09%). Additionally, by adding non model crops to the training dataset, we improved accuracy for these crops, demonstrating the model’s scalability. This model has significant potential for application in smart farming, particularly for Solanaceae crops, and could be adapted to other crop varieties by expanding the training dataset.