Smote-DL: A Deep Learning Based Plant Disease Detection Method
Subham Divakar, Abhishek Bhattacharjee, Rojalina Priyadarshini
Abstract
In the due course of time, computer vision, machine learning and deep learning has been widely used to detect disease in the plant leaf. Most works done in this area focuses upon coming up with accurate models but does not focus on the false predictions which could be a serious cause. Misdiagnosis of the plant leaf could cause large scale crop destruction. We used a publicly available dataset which contained four categories of images belonging to Apple Plant-Healthy, Scab, Rust and Multiple disease.However this dataset upon visualization was found to be imbalanced. Our main objective isto reduce the false predictions. The main contribution lies in the use of SMOTE method to balance the dataset and the novel Ensemble algorithm which uses both F1 score and accuracy to compare and come up with the best classifier from among the classifiers. Upon experimentation we came up with Efficient NetB7 as the best classifier from our list of classifier which had both good accuracy and good F1 Score. It also predicts whether a leaf image has multiple disease or not which helps to reduce false predictions further.