A more efficient process for asphalt maintenance through automation.

The customer
Asfalt-onderhoud is a company that does small-scale maintenance on asphalt. They repair cracks and holes in roads, bicycle paths, footpaths and playgrounds for a number of Dutch provinces and municipalities. This is a very demanding 24/7 job with strict standards.

The challenge
The total repair time determines their billable costs. Once the maintenance is completed, a car with a mounted camera drives over the repaired section to take pictures of it. Then someone inspects the recording and manually checks off all the repairs in the recorded video. The results are then exported to a spreadsheet to create an overview and finally an invoice.

The solution
We created a scalable cloud service that detects damage and repairs in the camera images using computer vision and deep learning. Detected repairs can be immediately inspected as an overlay on the camera images and easily exported.

The result
Our solution largely automates the time-consuming and costly task of annotating the recordings. It speeds up this process enormously so that Asfalt-onderhoud can work far more efficiently. The job is done faster. With less time spent on peripheral issues such as administration, more can be devoted to the maintenance itself.

The results of this collaboration are clearly presented in this short video.