Plant disease detection is essential in precision agriculture to safeguard crop productivity and prevent economic losses. Traditional disease diagnosis relies on manual visual inspection by experts, making the process subjective, time-consuming, and inaccessible to smallholder farmers. This research proposes a deep learning-based approach that utilizes convolutional neural networks (CNNs) to automatically classify plant leaf diseases from images. The PlantVillage dataset, comprising over 50,000 samples of healthy and diseased leaves, is used to train and validate the system. The methodology includes image preprocessing, data augmentation for generalization, hierarchical feature extraction, and softmax-based multiclass classification. Experimental results show that the model achieves an accuracy of 98.27% and demonstrates robust performance across various crop disease categories. The proposed solution can be easily integrated into smartphone applications, enabling real-time disease detection and timely guidance for farmers. This system contributes significantly to sustainable agriculture by reducing the misuse of pesticides and improving crop yields.
A 2025 study studied this question.