YOLOv8 Algorithm Implementation for Detecting and Classifying Waste Types Using CCTV Cameras

Abstract

Waste accumulation has emerged as a critical global environmental concern, prompting the need for intelligent and automated waste management solutions. This study presents the development of a real-time waste detection and classification system using closed-circuit television (CCTV) cameras integrated with the You Only Look Once version 8 (YOLOv8) deep learning algorithm. The dataset used in this research was collected from two main sources: direct image captures from various physical locations and publicly available waste image datasets from the Kaggle platform. All images underwent preprocessing steps including resizing to 640×640 pixels, normalization, and annotation to ensure data consistency and quality. To enhance the model’s robustness, data augmentation techniques such as horizontal flipping, cropping, rotation, brightness adjustment, and mosaic transformation were applied. The dataset was split into 80% for training and 20% for testing. The YOLOv8 model was trained with optimized hyperparameters suitable for object detection tasks. Performance evaluation was conducted using standard metrics including precision, recall, and mean average precision (mAP). The model achieved a precision of 0.486, recall of 0.361, mAP@50 of 0.368, and mAP@50–95 of 0.273. The detection accuracy varied across different waste categories, with higher performance observed for aluminum cans and cardboard, while lower performance was noted for aerosols and aluminum caps. These results indicate that YOLOv8 shows potential for waste classification but still requires improvements. Future work should focus on expanding dataset diversity, refining data augmentation strategies, and further hyperparameter tuning. This study contributes to the advancement of intelligent waste management through the application of deep learning and computer vision.