Prediction and Diagnosis of Pomegranate Fruit Disease Using CNN by Visual Field Data

  • Jayashri Patil, Sachin Naik

Abstract

Fruits are heavily affected by climatic conditions that further results in reduced agricultural yield. Due to this, economy of agriculture gets affected. In addition to this, condition becomes even worst when fruits are infected by any disease. Hence, modern agricultural techniques and systems are required to detect and prevent the fruits from being affected by different diseases. In this paper, a technique is suggested that will help the farmers to identify fruit disease. In this technique, the system consists of trained data set of images for the pomegranate fruit and leaf. User gives an input image that is processed through various components to detect the severity of disease by comparing with trained dataset of images. In the proposed method HSV color model is used to detect the leaf and pomegranate fruit from background. Image segmentation methods are applied for further processing and extracting region of interest (here leaf or pomegranate fruit area). This cropped region of interest image patches are used as database with label of disease to train the neural network model. Once the network is trained a random image is taken as input, here again the process of region of interest identification (here leaf or pomegranate fruit) is done and then fed to trained neural network model. It gives output the matching results related to each disease type. The matching score which has highest numerical figure is assumed as fruit disease of that type. The accuracy varies depending on the number of epochs used to train the neural network model, size of image database and also type of random input image selected for prediction of disease. For the proposed method developed we achieve accuracy of more than 95% in random samples of input images.

Published
2021-10-24
How to Cite
Jayashri Patil, Sachin Naik. (2021). Prediction and Diagnosis of Pomegranate Fruit Disease Using CNN by Visual Field Data. Design Engineering, 6773 - 6788. Retrieved from http://thedesignengineering.com/index.php/DE/article/view/5663
Section
Articles