Adoption of Artificial Intelligence (AI) Technology

Road Defect Detection System (RDDS)

The Highways Department (HyD) developed the groundbreaking Road Defect Detection System (RDDS) in 2024. This state-of-the-art system redefines road maintenance efficiency by harnessing the power of GeoAI to fuse Artificial Intelligence, geospatial data as well as photogrammetry technology to automatically detect road cracks and discoloured road markings from road images captured by a vehicle-based camera system.

RDDS is specifically trained on Hong Kong’s unique road conditions to ensure the detection of road defects with high relevance and precision. It is also capable of generating comprehensive defect reports that include the three-dimensional coordinates, dimensions and affected area of detected defects to facilitate efficient planning of repair works.

Building on this success, HyD is continuously advancing the system through a visionary upgrade. The next-generation evolution is currently in progress and expands from "Convolutional Neural Networks" (CNN) to the integration of the more advanced "Vision Transformer", catapulting the image recognition capabilities to new heights of processing speed while a wider range of defect recognition. Beyond road surfaces, this expanded intelligence will proactively monitor diverse road assets from identifying faded/ bended/ obstructed traffic signs, unwanted vegetation on carriageways, as well as defective manhole covers/ gully gratings and traffic bollards. The next RDDS is scheduled for full production by 2027.