Rachnova Intelligence
HunterRepo
A licence-plate recognition app for vehicle repossession teams, powered by a YOLO computer-vision model on iOS and Android.
Overview
Repossession agents spend their shifts scanning streets and car parks for specific vehicles. HunterRepo turns the phone camera into that search: a trained YOLO model finds licence plates in view, the app reads them into text, and agents can match vehicles without typing a single plate by hand.
The challenge
Agents work from moving cars, at night and in all weather, and plates appear at every angle and distance. Recognition had to stay reliable in those conditions and fast enough on a phone that it keeps up with the drive, not just score well on a test set.
- Practice
- Rachnova Intelligence
- Platforms
- iOS, Android
- Scope
- Computer vision, Model training, Mobile engineering
What we built
Inside HunterRepo.
- 01
Plate detection model
A YOLO object-detection model trained and evaluated on licence-plate imagery to locate plates in each frame.
- 02
Plate reading
Detected plates converted to text, ready to match against the vehicles a team is assigned to recover.
- 03
Detection pipeline
Pre-processing, inference and post-processing, including confidence thresholds and non-maximum suppression, wired end to end.
- 04
Field-ready apps
iOS and Android apps with live camera capture and results drawn on screen, built for one-handed use on the road.
At a glance
- YOLOobject-detection model
- Detect + readplates located and converted to text
- 2platforms: iOS and Android
Technology
- YOLO
- Computer vision
- OCR
- iOS
- Android
Delivered through our Applied AI service.