LeafYogi: Deep Learning based Mobile Application Framework for Ayurvedic Leaf Classification

Indian forest is the principal depository for useful medicinal herbs. Medical plants along with leaves have medical values and helps in curing many diseases. But correctly identifying leaves is laborious that takes lots of time and expertise. Therefore, the identification of these medical species is a challenging task. As a result, vision based approach helps the regular people in rapid identification of herbal leaves. This paper highlights vision based system to recognise herbal leaf by creating advanced deep learning models such as YOLOV5, YOLOV8 and Resnet-50. The Dataset for Indian Medicinal Plant Species Analysis and Recognition (DIMPSAR) which was published in 2023 is used to evaluate deep learning models. The experiment is performed on Colab environment. The evaluation metrics showed that YOLOV8 outperforms the other two models in term of accuracy with 99.2 percentage. The best architecture is further deployed in mobile application using teachable machine.