Development Of An Improved Tomato Leaf Disease Detection And Classification Method

Abstract

Detection of the plant leaf diseases in earlier stage is beneficial for Indian Economy. The study shows the 10-30% of crops are damaged due to diseases, which is not detected in curing stage. Different leaf disease detection methods are used for different crops. The pretrained Deep Learning Model is used to detect and classify the Tomato Leaf diseases. Dataset of the Tomato Leaf Images is collected from plant village repository. It is divided into categories, six diseased and one healthy. The implementation is done in MATLAB®. Features are extracted from the Feature Layer of the Pre-trained model of ResNet i.e. Fully Connected Layer. It is used to train the model for tomato leaf dataset. The training and testing are defined in a separate phase. The classification is done by the linear learner of the ECOC. It returns a pool trained multiclass error correcting model. To evaluate the trained model various parameters are calculated. The proposed model is able to classify the diseases has a higher Accuracy, Precision, F -Score, Specificity and False Positive Rate. The results of trained model are found to be more accurate than base article.