Analysis on Cassava leaf disease prediction using pre-trained models

Cassava plant is the most widely planted cash and food crop around the world. It has various mineral contents and vitamins so it is taken as a nutritive supplement food in the period of drought. The cassava plant production is very easy with the tropical climate but with the sudden change in the climate level and watering issues there may be a infection caused by fungus. Inorder to identify the level and variety of fungal or viral disease manual examination have to be made. However, this is time-consuming, costly, and lacks the knowledge to recognize the disease at the proper time. This leads the farmers to apply preventive strategies to non-infected cassava plants and affects the yields. This have been done to identify and categorize few kinds of disease infected on the cassava plant as it will be easy for the farm workers to take a preventive measure on time. This work mainly focuses on three different viral diseases of cassava plant namely Cassava Bacterial Blight, Cassava Mosaic Disease, Cassava Green Mite, Cassava Brown streak Disease and a healthy plant in order to expedite disease identification. In this project, there are 21,375 images of Cassava plants that will be used to train a disease classification model and we give different images to test the model using convolutional neural network (CNN). The model has been trained with two different CNN model architecture called EfficientNetB4 and Inception-Resnet-V2. The accuracy have been calculated and analysed that Inception-Resnet-V2 outperforms other model.