Research Article
Medical Plants Identification Using Leaves Based on Convolutional Neural Networks
@INPROCEEDINGS{10.1007/978-3-031-48888-7_14, author={B Ch S N L S Sai Baba and Mudhindi Swathi and Kompella Bhargava Kiran and B. R. Bharathi and Venkata Durgarao Matta and CH. Lakshmi Veenadhari}, title={Medical Plants Identification Using Leaves Based on Convolutional Neural Networks}, proceedings={Cognitive Computing and Cyber Physical Systems. 4th EAI International Conference, IC4S 2023, Bhimavaram, Andhra Pradesh, India, August 4-6, 2023, Proceedings, Part I}, proceedings_a={IC4S}, year={2024}, month={1}, keywords={Image Processing CNN Medical Plants Computer Vision}, doi={10.1007/978-3-031-48888-7_14} }- B Ch S N L S Sai Baba
Mudhindi Swathi
Kompella Bhargava Kiran
B. R. Bharathi
Venkata Durgarao Matta
CH. Lakshmi Veenadhari
Year: 2024
Medical Plants Identification Using Leaves Based on Convolutional Neural Networks
IC4S
Springer
DOI: 10.1007/978-3-031-48888-7_14
Abstract
The ayurvedic medicines have played a crucial role in health system, only a few experts could identify the herbs and know the ayurvedic properties of these herbs. These medicines prepared from the herbs having less side effects as compared to other general medicines. Most of the patients and general medicine users with different diseases are not unaware of the existence of herbal plants and their medical uses and benefits. To make ease of identifying the plants and its medical properties based on the leaf structure, authors developed a system having three architectures which works with Convolutional Neural Networks. Resnet-18, Resnet-50, MobileNet-V2 architectures were used in freeze and unfreeze layers settings. Authors considered ten different kinds of herbal leaves for implementation of the system, in which two thirds of the data used for training and one third for testing. The overall performance of this architecture is checked using accuracy measure and it is observed that three models with freeze layers were showing good performance. Out of these three architectures, Resnet-50 shown accuracy of 95.33%.
