Research Article
Classification of Herbal Plants Based on Leaf Images using Convolutional Neural Network
@INPROCEEDINGS{10.4108/eai.11-10-2022.2325271, author={Kana Saputra S and Debi Yandra Niska and Insan Taufik and Mhd Hidayat and Dinda Farahdilla Dharma}, title={Classification of Herbal Plants Based on Leaf Images using Convolutional Neural Network}, proceedings={Proceedings of the 4th International Conference on Innovation in Education, Science and Culture, ICIESC 2022, 11 October 2022, Medan, Indonesia}, publisher={EAI}, proceedings_a={ICIESC}, year={2022}, month={12}, keywords={convolutional neural network herbal plants leaf images}, doi={10.4108/eai.11-10-2022.2325271} }- Kana Saputra S
Debi Yandra Niska
Insan Taufik
Mhd Hidayat
Dinda Farahdilla Dharma
Year: 2022
Classification of Herbal Plants Based on Leaf Images using Convolutional Neural Network
ICIESC
EAI
DOI: 10.4108/eai.11-10-2022.2325271
Abstract
Indonesia is an agricultural country that is famous for its wealth of spices and herbal plants. Herbal plants themselves have thousands of species. There are 40,000 species of herbal plants that have been known in the world, and around 30,000 species to be in Indonesia. Herbal plants are a source of new active compounds that have pharmacological and therapeutic effects, both when used directly and through various extraction processes. Herbal plants can be distinguished from the shape of the leaves because each type of plant has different leaf features. Laboratory-based testing also requires skills in sample processing and data interpretation, in addition to time-consuming procedures. Therefore, a simple and reliable herbal plant recognition technique is needed to quickly identify herbs, especially for those who are unable to use expensive analytical instrumentation. This study aims to identify types of herbal plants based on leaf images quickly and accurately using the Convolutional Neural Network method which is part of Deep Learning. This study uses several architectural models of Convolutional Neural Network to classify types of herbal plants. The best accuracy value with the VGG16 architecture is 90% with 93% precision, 90% recall, and 90% F-measure. The VGG16 architecture used epoch = 20, batchsize = 32, and validationsplit = 0.2. The result show that CNN Algorithm with the VGG16 architecture is able to classify types of herbal plants with good accuracy.
