Classification of Stinging Nettle Plants Based on Leaf Images Using the CNN Method (Case Study: Biru-Biru Village)

Abstract

The high number of skin irritation cases among residents in Desa Biru-Biru due to direct contact with stinging nettle plants highlights the need for an automatic identification system to distinguish plant types. This study aims to develop a leaf image classification model for stinging nettle plants using the Convolutional Neural Network (CNN) algorithm with the MobileNetV2 architecture. The image dataset was collected directly from the study area and classified into four categories: Jelatang Ayam, Jelatang Gajah, Jelatang Niru, and Non-Nettle plants. The research stages include data collection and analysis, pre-processing (resizing, normalization, augmentation), data splitting (70:10:20), model training, performance evaluation (accuracy, precision, recall, and F1-score), and web-based system implementation. The test results show that the model achieved an accuracy of 98%, with the highest precision score of 0.98, recall score of 0.98, and F1-score of 0.98. The system has also been successfully implemented as an interactive web application that allows users to identify nettle plant types quickly and accurately. This research contributes to risk mitigation efforts related to harmful plants in rural environments through the application of digital image processing technology.