Classification of Diseases in Vegetable Bearing Plants Using Deep Learning Methods

Deep Learning methodologies can be used for computer vision application. Early detection of disease in crops is critical for producing profitable crop yield. To detect diseases in tomato and potato plant leaves, a convolution neural network-based classifier was created in this work. The types of diseases that are detected are of fungal, bacterial, viral and due to pests. The dataset used was an augmented version of Plant-Village Dataset. The proposed model achieved 98.14% training accuracy and a validation accuracy of 96.8 % when only potato plant data was used. The proposed model correctly classified with an accuracy of 93% in training and an accuracy of 91 % in validation. When the models were tested with new images in varying lighting conditions, they achieved an average accuracy of 85.3%.