An Effective Deep Learning Approach to Plant Disease Detection

Plant diseases are considered a major problemthat has an immense impact on agriculture and the economy.Early detection of plant diseases can result in less crop lossand significantly decrease financial losses. This study identifiesplant diseases using deep learning (DL) and machine learning(ML) methods with plant leaf images. DL methods such asDenseNet121, DenseNet169, DenseNet201, and VGG19 are used.Aditionally, ML methods including Support Vector Machine(SVM), Random Forest (RF), and Logistic Regression (LR) havebeen implemented. The plant leaf images are collected fromthe PlantVillage dataset. DenseNet201 with SVM appears as theoutperformer and provides the best results of 100% accuracywith precision, recall, F1-score, and AUC. Index Terms—Plant Disease, Deep Learning, Machine Learn-ing, Feature extraction, Classifier, Recognition