Detection of Solid Waste Deposits in Urban Areas using Artificial Intelligence and Image Processing: a Literature Review

Abstract One of the main problems that governments around the world face is the prevalent presence of solid wastes in their countries. Scattered solid wastes in rural and urban areas cause serious problems to people and the environment. In the last years, different solutions for solid waste management have been developed using technology and artificial intelligence. Computer vision is one of these areas in constant development, with improvements in techniques and algorithms to detect and classify objects in images and videos. By doing a literature review in different databases, we found 17 studies from IEEE, ACM, Science Direct, Springer and IOPScience that address the use of artificial intelligence techniques to detect and classify solid waste deposits using computer images. We analyzed information about the object detection techniques and the dataset used for algorithm training in these studies. We also depicted the metrics used to evaluate the performance, accuracy, and precision to detect garbage on images. Deep learning is the main technique used for image processing. YOLO, Deep CNN and Faster R-CNN are the principal techniques used for classification and detection of solid waste due to their speed and accuracy. These results may be very useful to induce and to guide the development of tools to detect solid waste in our country.