Early Identification of Disease in Indian Gooseberry using Machine Learning and Deep Learning Techniques

The Gooseberry cultivation is crucial for both commercial and domestic purposes, but its productivity is threatened by various diseases. Some of the commonly occurring disease are soft rot and rust disease. Traditional methods for disease detection and classification consume a lot of time and often lack accuracy. In this study, we design a novel approach leveraging machine learning as well as deep learning techniques, incorporating Support Vector Machine, Convolutional Neural Networks and ResNet18 for the prediction of gooseberry fruit diseases. Furthermore, we integrate image segmentation using the K Means clustering technique to precisely identify regions of interest within the images. Our methodology involves preprocessing the images, applying image segmentation to the preprocessed images, training the deep learning models on annotated datasets, and validating the models using testing datasets. We specifically employ K Means clustering for image segmentation to isolate disease affected areas. Experimental results show the efficacy of the proposed approach in accurately predicting gooseberry fruit diseases and segmenting the affected regions within the images. It was observed that ResNet18 was found to predict the disease in the Gooseberry fruit precisely with an accuracy of 96.1%.