We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
A Comparative CNN Based Deep Learning Model Investigation for Identifying and Classifying the Leaf Diseases of Arachis Hypogea (Groundnut Crop) Grown in The Semi-Arid Landscapes of Villupuram District of Tamil Nādu.
- Authors
S., Sivasankaran; Mohan, K. Jagan; Nazer, G. Mohammed
- Abstract
Groundnuts are India's most important oil seed crop, and they serve a critical part in bridging the country's vegetable oil shortfall. Groundnut cultivation has been hampered by a variety of fungal, viral, and bacterial infections resulting in significant production loss. Adapting deep learning Investigations towards the identification and classification of various leaf diseases of groundnut may enroute a precisive model to combat such leaf diseases in its early stages itself. In this research article, we have considered the major leaf diseases which affects the groundnuts crops cultivated in the semi-arid geographical landscapes of Villupuram district in the state of Tamilnadu. We have made a comprehensive investigation based on the collected groundnut leaf images in a total of (1950) both diseased and healthy from the various groundnut fields of Villupuram, and trained a deep convolutional neural network based on VGG 16 Model neural network with the utilization of Adam optimizer to identify 5 major groundnut leaf disease classes along with the healthy leaf class. The trained model has achieved an overall accuracy of 99.82 %. This investigational model has been compared for its performance and error measures in par with the CNN based VGG16 model with RMSprop optimizer, Alexnet and Inception_V3 models and have proved to be outperforming all the compared baseline models in terms of performance metrics.
- Subjects
TAMIL Nadu (India); DEEP learning; LEAF diseases &; pests; PEANUT diseases &; pests; ARID regions; CONVOLUTIONAL neural networks
- Publication
Journal of Algebraic Statistics, 2022, Vol 13, Issue 2, p232
- ISSN
1309-3452
- Publication type
Article