A Deep Neural Network based disease detection scheme for Citrus fruits

One of the most significant factors is the quality evaluation of agricultural products in supporting their marketability and controlling waste management. To classify the fruits into healthy and defected class, deep learning algorithms have been implemented to perform citrus disease detection. This study aims to use the dense CNN algorithm to detect and provide an effective method for detecting the apparent defects of citrus fruit. Citrus fruit images are collected and put in two classes of good and damaged ones, to recognize and categorize the image dataset. Firstly, a dense CNN model was used without doing preprocessing and data augmentation on 150 images and achieved an accuracy of 67 percent but the proposed model has used data augmentation and pre-processing to enhance the CNN performance and have used 1200 images. Further, the proposed model is compared with the dense model where data augmentation and pre-processing techniques have not been used. The overall accuracy of the proposed model is 89.1%. The results show that techniques of data augmentation and preprocessing have delivered promising insights to estimate citrus fruit's damages.