A Novel Herbal Leaf Identification and Authentication Using Deep Learning Neural Network
Haryono, Khairul Anam, Azmi Saleh
Abstract
Herbal plants are plants that can be used as an alternative to the natural healing of diseases. The existence of herbal plants is still not widely known by the public. It is due to many types of medicinal plants so it requires special knowledge to recognize them. A smart and accurate herbal leaf recognition system is needed to overcome this. This study aims to identify and authenticate herbal leaves using the convolutional neural network and Long Short-Term Memory (CNN-LSTM) methods. Identification was carried out on nine types of herbal leaves divided into two-thirds of training data and one-third of testing data. The results of the identification process were validated by other data not included in training data and testing data, as well as leaf data other than the nine types of leaves identified. The CNN-LSTM method shows good results in the identification process, with an accuracy of 94.96%.