With rapid urbanization, maintaining urban infrastructure has grown into a gigantic requirement. Proper and timely identification of infrastructure assets, such as manhole covers and drainage, is of utmost importance to ensure that water drainage and sewerage systems work properly within the precincts of a city. The classical methods of inspection have contributed to being slow, expensive, and full of errors. The paper tries to implement the use of YOLO in the automatic detection of manhole covers and drainage in images derived from Google Street View. This study will be focused on how to integrate results from object detection with MIS in order to monitor city infrastructures and optimize the planning of maintenance. These results proved that YOLOv11 has a very high accuracy rate and has identified manhole covers and drainage from imagery on Google Street View. Performance metrics included mAP@0.5 and mAP@0.5-0.95, which described sensitivity and accuracy of the model, while the FPS analysis described the applicability in real time. Those kinds of findings have underlined that AI-based solution usage is efficient in the automatic monitoring and management of urban infrastructure and prove their potential to contribute much to decision support systems.
Read full abstract