Trupti
@kjei.edu.in
Assistant Professor,E&TC
Trinity Academy of Engineering,Pune
RESEARCH INTERESTS
Biomedical Signal Processing,IoT,Microcontroller
Scopus Publications
Scholar Citations
Scholar h-index
Scholar i10-index
Scopus Publications
Mayuresh B. Gulame, Trupti. G. Thite, and Kranti D. Patil
IOP Publishing
Abstract A vital sector of India’s economy is agriculture. Identification of plant infections is crucial to preventing crop damage and further disease. The majority of plants, such as apple, tomato, cherry, and grapes, have leaves that appear to have disease signs. The plant health can be monitored through images to precisely predict the disease and to take early preventative action. The traditional method is to manually inspect the plant leaf to identify the kind of disease, as done by farmers or plant pathologists. In this research, we presented a deep CNN model termed as Decompose, Transfer, and Compose (DTComp) for the classification of plant disease. The deep learning model makes predictions more quickly and precisely than manual plant leaf observation. Out of all the pretrained deep models, the ResNet50 model achieves the highest accuracy for classification. DTComp can handle any anomalies in the images using class decomposition approach to examine the class boundaries. The experimental findings demonstrated DTComp capacity for detecting plant disease instances on dataset gathered from multiple villages using the Kaggel Open Source platform. DTComp can successfully identify plant disease with a high accuracy of 98.30% from images. Additionally, this model can be deployable on real-time systems equipped with a Raspberry Pi and a camera module.
Trupti G. Thite and Daulappa G. Bhalke
Inderscience Publishers
Trupti G. Thite and D. G. Bhalke
Springer Nature Singapore