Fruit Disease Classification using Convolutional Neural Network

Abstract

Convolutional Neural Network (CNN) demonstrates good success rates in image classification and produces accurate precision values. This project aims to create a comprehensive in-depth reading model that identifies the category of fruits from the images. By use a data-enhancing method that increases the number of training data by creating new images using available images to train the model with multiple images and improve the chances of separating new images. By dividing the accuracy of the image and obtaining the percentage of accuracy of the loss as a separation of diseases. Using a category filter with better accuracy is achieved than the previous methods. The problem with these various algorithms is that the accuracy is reduced as certain sources of error have not been removed. In-depth Learning brings methods, methods, and practices that can help solve critical analysis and predictability. Diagnosis of fruit diseases has also been a difficult task to achieve through the use of computers. Using the new development of neural model networks a relatively good image classification of new and sick images was created. The database is available at Kaggle.com which contains images of three different fruits but images of oranges were used in this experiment to make the model clearer. Many dissimilar types of fruit will be need of the model to identify the type of fruit which increases the cost of calculation. Instead of making separate models allows the model to specify and increase accuracy in a single domain, this will usually allow for better results.