Application Development for Image Recognition of Herb Flowers Using Deep Learning

Abstract

This study presents the development of a mobile application for herbal flower recognition using deep learning and image processing techniques. The application was designed to classify and provide information about the medicinal properties of herbal flowers, particularly in the treatment and alleviation of various health conditions. The classification model was trained on a dataset comprising 200 images distributed across ten species of medicinal flowers (Hibiscus, Plumeria, Damask Rose, Sunflower, Bougainvillea, Marigold, Sacred Lotus, Butterfly Pea, Jasmine, and Yellow Elder) using the Teachable Machine platform. The application was implemented in Dart programming language.

Experimental results demonstrated that 1) the model achieved 93.00% prediction accuracy in herbal flower classification tasks; 2) the application was rated at a good level in terms of technical performance and ease of use by domain experts; and 3) user satisfaction surveys revealed exceptionally high approval ratings. The developed application has demonstrated potential in promoting learning and disseminating knowledge about Thai herbal plants. This technology enables users to capture flower images of flowers, after which the system identifies the specimen and provides comprehensive information about its medicinal properties and traditional usage.