Image-based Rice Leaf Disease Identification Using GoogleNet Transfer Learning Model

  • P.Rajasekar, J.Arputha Vijaya Selvi
Keywords: Rice plant disease identification, Deep learning, Convolutional neural network, Transfer learning, GoogleNet, Image classification

Abstract

The quality and quantity of agricultural outcome is affected by various plant diseases. Plant diseases act as a threat to food safety and production. In this digital era, we have to identify and classify plant diseases to achieve high productivity.  Manual identification of plant diseases by farmers is very difficult due their limited awareness on the types of diseases. Recently, many techniques are proposed to solve this issue in which deep learning is most preferred due to its incomparable performance. In this paper, the transfer learning-based deep convolutional neural network called GoogleNet is proposed to identify and classify the rice plant diseases. Since, the data set on rice leaf is limited, this research work created a large set of dataset by the image-augmentation process. A pre-trained model on rice leaf disease is used by the proposed GoogleNet transfer learning technique. Three rice leaf diseases are taken (Bacterial leaf blight, Brown spot, and Leaf smut) to be identified and classified. The framework proposed in this paper gives a great improvement in disease detection performance compared to other state-of-the-art methods. Over 92% of average accuracy in detection and classification of rice leaf disease is achieved for the data set utilized. 

Published
2021-11-08
How to Cite
J.Arputha Vijaya Selvi, P. (2021). Image-based Rice Leaf Disease Identification Using GoogleNet Transfer Learning Model. Design Engineering, 10325-10333. Retrieved from http://thedesignengineering.com/index.php/DE/article/view/6096
Section
Articles