Abstract: Nowadays, we are facing many problems of environmental pollution. One of them is the process of waste management since the amount of waste is proportional to the number of people in urban areas. The classification of waste plays an important role in the recycling of waste contributing to minimizing the risk of spreading pathogens, toxic and dangerous elements. We are in the fourth industrial revolution, applying cutting-edge technology is the trend, specifically deep learning techniques in the waste recycling process. Smart waste recognition also contributes to saving human resources and reducing costs for waste collection and recycling. In this paper, we propose a waste detection and classification model based on YOLOv4 architecture. We experimented and obtained mAP of 90.27%, F1-score of 86% on the dataset that we synthesized including 4 main types of waste: plastic, metal, glass, and paper.

Keywords: Computer Vision, Object Detection, Classification, Deep Learning, YOLOv4.

PDF | DOI: 10.17148/IJARCCE.2022.11503

Open chat
Chat with IJARCCE