Title: The effect of plant leaf disease on environment and detection of disease using convolutional neural network

Authors: Shivendra Kumar Pandey; Sharad Verma; Prince Rajpoot; Rohit Kumar Sachan; Kumkum Dubey; Neetu Verma; Amit Kumar Rai; Vikas Patel; Amit Kumar Pandey; Vishal Singh Chandel; Digvijay Pandey

Addresses: Department of Information Technology, REC Ambedkar Nagar, Ambedkar Nagar, India ' Department of Information Technology, REC Ambedkar Nagar, Ambedkar Nagar, India ' Department of Information Technology, REC Ambedkar Nagar, Ambedkar Nagar, India ' Computer Science Engineering and Technology, Bennet University, Greater Noida, India ' Department of Computer Science, United University Prayagraj, Prayagraj, India ' Department of Computer Science, MNNIT Allahabad, Prayagraj, India ' Department of Civil Engineering, REC Ambedkar Nagar, Ambedkar Nagar, India ' Department of Electrical Engineering, REC Ambedkar Nagar, Ambedkar Nagar, India ' Department of Applied Science and Humanities, REC Ambedkar Nagar, Ambedkar Nagar, India ' Department of Applied Science and Humanities, REC Ambedkar Nagar, Ambedkar Nagar, India ' Department of Technical Education, IET, Lucknow, India

Abstract: Agriculture significantly influences people's daily lives and financial well-being, making it crucial to enhance crop productivity. Recognising and preventing plant diseases are vital in achieving this goal, as diseases can severely impact production and the environment. Traditional methods of disease detection, relying on human visual inspection, are time-consuming, expensive, and prone to errors. In contrast, employing image processing techniques and convolutional neural networks (CNNs) can offer fast and accurate results. This paper compares deep CNN, VGG19, and ResNet, to detect plant diseases from leaves images. We train these models on a dataset containing images of newly discovered plant diseases. The results demonstrate that the ResNet model outperforms the customised deep CNN and VGG models, achieving an impressive accuracy of 99.3% with the fewest parameters.

Keywords: convolutional neural network; CNN; deep learning; ResNet; VGG19.

DOI: 10.1504/IJGW.2024.138128

International Journal of Global Warming, 2024 Vol.33 No.1, pp.92 - 106

Received: 28 Jun 2023
Accepted: 21 Aug 2023
Published online: 29 Apr 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article