Volume 18, No. 6, 2021

Comparative Disease Prediction Performance Analysis For Indian Brinjal Plant

Balwant Gorad , Sirbi Kotrappa


As per NCRB Government of India’s report on farmers, around 35 famers die every day. There are many reasons for this. Some of the major causes are low crop production after many efforts, increased cost of cultivation and management, debt on farmer, low crop prices and many other environment issues. Brinjal is one of most cultivated crop in major part of India. One of the major reasons of above issues is improper disease prediction and management on the crop. This paper presents comparative analysis of disease prediction on Indian Brinjal plant using various deep learning techniques. Deep learning has performed outstandingly in image classification problems. Various deep convolutional neural network (DCNN) models include VGG16, VGG-19, ResNet50, ResNet101, InceptionV3 and Dense Net 121 are implemented to predict the disease on Brinjal plant. The high-quality preprocessed dataset collected from real field with data augmentation of 40,336 images is used to conduct this research work. These implemented models are achieved the training accuracy of 90.92%, 90.69%, 45.80%, 52.21%, 89.17% and 92.35% respectively whereas validation accuracy of 97.01%, 99.66%, 45.92%, 55.86%, 95.10% and 97.89% respectively. These accuracy results showed that the DenseNet121, VGG19, VGG16 and InceptionV3 deep convolutional neural network models are promising way for effective implementation of such disease prediction systems in real time agricultural field.

Pages: 2087-2107

Keywords: Deep Learning, Convolutional Neural Network, Disease prediction, Agriculture

Full Text