Real-time detection of construction and demolition waste impurities using the improved YOLO-V7 network

Construction and demolition waste accounts for a considerable part of the total waste flow of the city. The most common way to recycle it is to make it into recycled aggregate. In the process of recycling and preparing recycled aggregate from the construction and demolition waste, it is necessary to manually screen out impurities that remain after wind selection, water floating, etc. This not only increases production costs but also affects the quality of recycled aggregates and the utilization rate of construction and demolition waste. This study proposes an automated method for detecting construction and demolition waste using an improved object detection network. By improving the feature fusion layer, the convolutional block, and the loss function of the YOLOv7 object detection network, the recognition accuracy, the recall rate, and the mean average precision of the network have been greatly improved, while the number of parameters has been further reduced. Therefore, the improved YOLOV7 network can effectively identify various impurities in the dismantled waste, providing technical support for automatic detection and screening of construction and demolition waste impurities robots, improving the efficiency of enterprise processing of construction and demolition waste, and indirectly alleviating environmental problems and resource waste caused by construction and demolition waste.