A Cutting-Edge Machine Learning Framework for Potato Leaf Disease Detection

Agriculture play important role in Indian economy and it provides livelihood to millions of people. Plant diseases are affecting this plant growth. If it is given timely treatment then it also works as boost of economy. The traditional methods of predicting plant diseases are extremely time-consuming and do not yield precise results. Machine learning based approach can automate the prediction of diseases in the plants. Objectives: Authors used four machine learning algorithms in the proposed work for the prediction of plant diseases, including K-Nearest Neighbor (KNN), Support Vector Machines (SVM), Decision Trees, and Convolutional Neural Networks (CNNs). Dataset is taken from kaggle which is having 4062 images which are used for training and testing of these models. The images are harvested from the Central Punjab region of Pakistan and depict different stages of potato leaf diseases typical in the region. Methods: Precision, recall, and F1 score are computed in order to assess the research work. The experimental results are verified by taking comparisons between the proposed models. Results: The outcome of the research work proved that CNN model performed well among the four models. For the CNN model Accuracy is calculated at 95%, Precision is 93%, Recall is 92% and F1 score is 92%.