<![CDATA[Indonesia is one of the world's megabiodiversity countries, possessing a rich diversity of herbal plants that play an important role in traditional medicine and biodiversity conservation. However, the identification of herbal plant species remains challenging for the public due to similarities in leaf morphology, increasing the risk of misidentification and inappropriate utilization. This study aims to develop a web-based identification system for Indonesian herbal plants using a Convolutional Neural Network (CNN) to facilitate accurate and accessible species recognition. The system employs the MobileNetV2 architecture to classify herbal plants based on leaf images and provides an interactive web interface for users. Experimental results showed that the proposed model achieved a training accuracy of 96.30% and a validation accuracy of 95.75%. Performance evaluation using 50 independent leaf images yielded an identification accuracy of 89.24%, with 47 images correctly classified. These findings indicate that the proposed system is capable of supporting reliable herbal plant identification and can serve as a can serve as a practical digital tool for supporting biodiversity monitoring, medicinal plant documentation, public education, and the sustainable utilization of Indonesia's medicinal plant resources.]]>
Copyrights © 2026
