Road Defect Detection
Road Defect Detection reviews video from cameras already fitted to council vehicles and flags surface defects for inspection and repair.
Keywords
- Tags: Computer vision
- Tags: Asset management
- Tags: Roads
- Stage: Production
- Type: Convolutional neural network
- Sector: Transport
- Language: English
How does our product work?
Footage is processed overnight. The model marks frames containing likely defects, estimates severity and attaches a location from the vehicle's GPS trace.
Every flagged defect is confirmed by an inspector before any repair is ordered. Faces and number plates are blurred before a human ever sees a frame.
Overview
Most road defects were previously found either by a resident reporting them or by a scheduled inspection that visits each road once a year. Both mean small defects become expensive ones before anyone acts.
Using vehicles that already drive these roads daily gives near-continuous coverage at almost no additional cost.
Owner and responsibility
More detailed information on the system
Here you can get acquainted with the information used by the system, the operating logic, and its governance in the areas that interest you.
System description
Detection runs on recorded footage, not in real time, and the vehicle operator has no role in and no visibility of it. Frames without a detection are discarded within 72 hours; frames with one are retained until the defect is closed.
References
Data protection impact assessment, including the decision to blur before human review.