Hi-Tech Plant Leaf Classification Using Artificial Intelligence (AI)

Abstract

This study was aimed at surveying wild fruit plant species in the Khok Phu Taka forest area and developing a system of plant leaf classification using AI technology (Polylthia debilis Finet & Gagnep and other plant species). Data were collected from the Khok Phu Taka forest, and the AI system (development of AI by training images (N=145)) was developed to classify the leaves of Polylthia debilis Finet & Gagnep and other plant species. The system accuracy was verified of 10% of the images by using YOLO v4. The data were analyzed by Confusion Matrix. The findings revealed that plant species in the study area include Polylthia debilis Finet & Gagnep, Polyalthia evecta (Pierre) Finet & Gagnep, Indian gooseberry (Phyllanthus emblica), and Ziziphus oenoplia (L.) Mill. var. oenoplia, etc. The data were then labeled to develop the system, and an accuracy of 93 % was achieved in classifying Polylthia debilis Finet & Gagnep leaves and other species. In conclusion, the developed AI system increases the efficiency of plant species identification through leaf classification, while it provides the method to identify various plant species in the future.