Potato plant leaf diseases detection and identification using convolutional neural networks

Abstract

Potato plant leaf diseases can cause significant crop losses and impact the quality of potato products. Early identification of these diseases can aid in timely intervention and treatment, minimizing the spread of diseases and reducing crop losses. In this work, we propose the incorporation of Convolutional Neural Networks (CNNs) for the automatic detection and identification of diseases present in potato plant leaves. We compiled a diverse dataset of images of potato plant leaves and used it to train and evaluate a CNN for the classification of potato leaf diseases. The results of the research work shows that the CNN was able to achieve high levels of precision score, F1 score, recall and accuracy over the training and testing splits of the dataset. Our findings demonstrate the potential of CNNs for the automatic identification of diseases present in potato plant leaves, and suggest that this approach could be applied to other types of plant diseases as well. The proposed method has the capability to improve crop management and disease control efforts by enabling the automatic and early detection of leaf diseases, which could have significant economic and environmental benefits. According to our result, the CNN offers a promising method for identifying and detecting leaf diseases on potato plants. This method may also be utilized to create an automated system for disease diagnostics in potato harvests. For the goal of identifying diseases in potato leaves, a Deep Learning (DL) base CNN model has been introduced in this research work. The proposed CNN technique has secured 98.1% accuracy rate for the Plant Village dataset, shows 3% improvement when compared to the existing methods.

Topics & Concepts