YOLO TrashNet: Garbage Detection in Video Streams

Cities today are beginning their transformation into "smart cities". Beside smart traffic, lighting, and energy management, smart waste is an integral part of any smart city. Particular attention should be given to the abandoned garbage both in public city areas or in places outside of town, such as countryside or suburban roads. Beyond causing the degradation of the area, abandoned garbage can cause pollution and have a negative impact on the quality of life of residents in these areas. Our work focuses on developing a software that is able to detect and report the presence of abandoned waste through the analysis of video streams in real-time. An improved YOLOv3 network model is adopted to perform garbage detection and recognition. The network has been fine-tuned on dataset collected for this purpose. The results show that the proposed approach may represent a major contribution for a more efficient waste management in smart cities.