Deep convolutional neural network-based cassava plant leaf disease classification

Abstract

Cassava is a woody plant in which leaf disease affects the productivity that primarily influences the farmers, as it occurs naturally. A Deep Convolutional Neural Network (DCNN) based leaf disease classification model is proposed. Automatic detection of cassava leaf disease assists humans in monitoring the huge farm of crops manually. Early detection of the disease symptoms on plant leaves aids in avoiding the spreading of leaf diseases throughout the farmland. The diseased leaf region is segmented. Its statistical features are extracted using Deep Learning (DL) technique and fed as input to the classifier with the training dataset model with various leaf diseases. First, the given input image is identified as a normal leaf or infected leaf. Followed by classifying the type of disease based on the pre-trained database with various datasets, namely Cassava Green Mite (CGM), Cassava Brown Streak Disease (CBSD), Cassava Bacterial Blight (CBB), Cassava Mosaic Disease (CMD) diseases.