Visual Detection of Waste using YOLOv8
Ram Bawankule, Vaishnavi Gaikwad, Indrayani Kulkarni, Shivam Kulkarni, Archana Jadhav, Nihar M. Ranjan
Abstract
With the increasing focus on environmental conservation, efficient and accurate waste classification has become a crucial task in the face of increasing urbanisation and industrialization. The challenges faced by waste management systems include pollution of water and air, the spread of diseases, inadequate infrastructure and resources, and the need for efficient waste segregation. Recent techniques proposed to address these challenges include imaging techniques, thermal analysis, and machine learning algorithms. However, these techniques have their limitations, such as specialized equipment and expertise to operate effectively, labour-intensive, posing risks to workers, and resulting in inadequate waste segregation. The proposed objective of this article is to evaluate the effectiveness of YOLOv8, the latest version of the YOLO series of object detection models, for automated waste sorting. The aim is to increase efficiency and safety in waste treatment processes by using YOLOv8 for automated waste sorting. The results demonstrate that YOLOv8 surpasses its state of art algorithms in detecting and classifying waste, making it a valuable tool for improving waste management practices.