Abstract:
Roads are considered as the lifeline of the economy of a nation. It is also considered as a metric for
comparing the economic standing of a nation on the international level. In order to maintain such
critical assets, a yearly maintenance plan is made. Implementing such plans and their inspection
requires a large amount of investment funds. While such funds are provided annually by a nation,
the costs incurred from such a maintenance plan can improved for economic efficiency, by
employing modern technologies. Traditional methods use manual methods of inspection, which are
time consuming, costly, prone to human error and requires a large workforce. Whereas modern
methods will replace such undesirable circumstances by automating the process of inspection via
computer vision techniques. Such large amounts of data will still require proper storage,
management, visualization and presentation. For this task, I-BIM is chosen which will store
detection data into a centralized system. It will also present data as objects in an I-BIM environment.
The entire process makes use of the YOLO v8 for object detection and Autodesk Civil 3D for an IBIM environment. The YOLO model was trained on two datasets, an open-dataset and a closeddataset. The model gave a mAP of 92% while the I-BIM model was developed from various
resources for different forms of data such as, Pavement thickness data taken from coring data of
past maintenance plans, geometric design data from national highway agency and GIS data from
open sources such as Google Earth and USGS. The integration is done via Dynamo which includes
the python language, where the crack detector itself is based in the python programming
environment. Thus, the data was integrated from the YOLO detector to Civil 3D with proper
georeferencing. While some limitations exist, the detector good results and integrated seamlessly
with the I-BIM model.
