Klasifikasi Sampah Organik dan anorganik Menggunakan Metode YOLOv5 Berbasis Roboflow

The waste problem in Indonesia continues to increase due to population growth and changes in consumption patterns. The lack of public awareness in separating organic and inorganic waste leads to ineffective waste management. This study aims to develop a classification system for organic and inorganic waste using the YOLOv5 method based on Roboflow. The dataset consists of more than 200 waste images that have been annotated and divided into 60% training data, 20% validation data, and 20% testing data. A pre-processing stage was conducted by resizing the images from 640×640 pixels to 320×320 pixels to improve training efficiency. The model was trained using the YOLOv5 architecture and evaluated using accuracy, precision, recall, and mean Average Precision (mAP) metrics. The experimental results show that the model achieved an average accuracy of 95% in classifying organic and inorganic waste. These findings indicate that the YOLOv5 method is effective for implementing a web-based waste detection and classification system and has the potential to support more efficient and sustainable waste management practices.