PLANT DISEASE DETECTION USING CONVOLUTIONAL NEURAL NETWORK

Abstract—Agriculture remains the backbone of many economies worldwide, yet plant diseases cause significant annual yield losses threatening global food security. Traditional disease detection relies on naked-eye observation by experts, a process that is time-consuming, subjective, and often inaccurate. This paper presents a deep learning-based approach for automatic plant disease detection using leaf images. A Convolutional Neural Network (CNN) was designed, trained, and fine-tuned on the publicly available PlantVillage dataset, which contains over 54,000 labeled images of healthy and diseased leaves across 14 crop species and 38 disease classes. The proposed system eliminates the need for manual feature extraction by learning relevant features directly from raw pixel data. Image preprocessing steps including resizing, normalization, and data augmentation were applied to enhance data quality. The CNN architecture comprises three convolutional blocks for hierarchical feature extraction, max-pooling layers for dimensionality reduction, and fully connected layers for classification. The model was trained using TensorFlow on a GPU-accelerated environment, achieving an overall classification accuracy of 96.77% on the validation set. Healthy leaves were never misclassified as diseased, and inference completes in under 0.5 seconds. This automated system can assist farmers, gardeners, and agricultural professionals in rapid and accurate disease diagnosis, enabling timely intervention, reducing pesticide overuse, and promoting sustainable agricultural practices.
Keywords - Plant Disease Detection, Convolutional Neural Network, Deep Learning, Image Classification, PlantVillage Dataset, TensorFlow, Precision Agriculture.