Using Convolutional Neural Network for Fruit Disease Detection in Machine Learning Approach

Abstract

Crop cultivation occupies a key place in the agricultural industry and is essential to supplying the world's food needs. The negative effects of diseased crops, which result in decreased production rates and resultant food loss, provide a serious concern. Machine learning techniques' quick development has created new opportunities for solving practical problems. In this study, we investigate the use of Convolutional Neural Networks (CNN) for fruit disease detection, utilizing the strength of machine learning techniques to automate fruit illness detection and diagnosis. This study uses machine learning to create a precise and effective system that can identify different fruit diseases for guava. We extract complex information from images, and CNN, a deep learning algorithm, promises to improve the precision of disease identification and classification. To do this, a sizable dataset containing pictures of both healthy and fruits with various ailments has been gathered. These pictures serve as instruction. We testing data for the CNN pre-trained InceptionV3, VGG16, and Resnet50 model. We aim to attain high overall accuracy in disease identification through intensive experimentation and model optimization. The findings of this study will help fruit disease detection systems develop by enabling early disease detection and facilitating quick action, thereby lowering crop losses. We also find the F1 score, confusion matrix, and accuracy. By offering effective and automated methods for fruit disease diagnosis, the suggested approach also illustrates the potential of machine learning techniques to revolutionize the agricultural economy.