Implementation of YOLOv5s for Automatic Waste Category Classification in Digital Waste Bank Systems
Implementation of YOLOv5s for Automatic Waste Category Classification in Digital Waste Bank Systems
Noorman Rinanto,Mat Syai’in,5 Authors,Harun Ismail
TLDR
It is demonstrated that YOLOv5s can accurately and efficiently classify waste objects, offering strong potential for integration into digital waste bank systems to enhance the efficiency and transparency of waste management processes.
Abstract
The increasing volume of organic waste in campuses or households demands innovative solutions for waste management and classification. This study proposes an automated classification system based on deep learning using the YOLOv5s algorithm to detect 14 categories of inorganic waste in real-time. The dataset consists of over 3.500 labeled images, annotated via Makesense.ai and augmented using Roboflow. The model was trained on Google Collaboratory for 100 epochs using the YOLOv5s architecture and evaluated based on precision, recall, F1-score, and mean Average Precision (mAP). Training result show [email protected] approaching 100% and [email protected]:0.95 around 85%, with an average confidence score of 88.30% during real-time testing using a webcam. These findings demonstrate that YOLOv5s can accurately and efficiently classify waste objects, offering strong potential for integration into digital waste bank systems to enhance the efficiency and transparency of waste management processes.