Identification of Selected Ethiopian Traditional Medicinal Plants Using Digital Image Processing and Deep Learning Techniques

Abstract

Habtamu Alebel, Million Meshesha, Wabi Jifara and Girma Asefa

Ethiopia is rich in biodiversity, and its vast landscape offers a variety of medicinal plants with significant cultural and therapeutic importance. The study is conducted to identify medicinal plants using image processing and deep learning techniques. This study explores both pre-trained models (such as VGG-16 and ResNet-50) and CNN sequential for creating models from scratch. Experimental results show that the Medicinal Plant Identification (MPI) model constructed from scratch performs better with test accuracy of 99.90% and an F-score of 99.00% performance as compared to pre-trained models such as VGG-16 and ResNet-50. The proposed model achieves high accuracy in image classification by utilizing multiple convolutional layers that capture fine- grained patterns in images.

PDF