AMSA
Road incidents detected with computer vision
Computer vision that analyses images captured on the road and gives maintenance a report sorted by priority.
Starting point
AMSA maintains roads and needs to detect incidents from images: damaged elements, obstacles, signage. Manual review was slow and hard to scale.
What we did
We are developing a computer vision system that analyses images captured on the road, detects and classifies incidents, places them on the map and prioritises them by severity. Maintenance teams get a report ready to act on.
The change
Before and now.
Before
- Driving the road, looking and taking notes
- Going back to the office and typing incidents into a report
- A slow review that is hard to scale
Now
- Cameras capture on every run
- AI detects, classifies and places every incident on the map
- The team heads out with a prioritised report
Step by step.
- Input
On-road images
The vehicle's cameras capture the road as it drives.
- AI
Computer vision
The model detects damaged elements, obstacles or signage and classifies them.
- Rule
Priority
Each incident gets a priority based on its severity.
- System
Incident map
Everything is placed on the map, with its image.
- Person
Maintenance
The team decides the route and where to start.
And up and running.
- Km 12.4Damaged signHigh
- Km 15.1VegetationMedium
- Km 18.7BarrierHigh
Result
Faster detection, less manual work and operational support that improves with every kilometre analysed.
Road teams head out with a clear list of where to act first.
Connected with
Related cases
16
modules in the recruitment platform
Headway
AI-Solutions
A custom ATS: 16 modules and a single database
A recruitment platform built from scratch that registers, searches and compares candidates with AI.