Hakatron
Mobile app using machine learning to recognize licence plates of stolen cars in real time.
Original state
HAKA operates social network groups and mobile apps enabling the general public to report stolen and found vehicles.
Solution
Increase the user engagement without the need of creating manual records.
Develop a real time image recognition of licence plates while driving and compare them on the fly with a database of stolen cars.
How We Got There
Through machine learning we created a library checking the licence plates on the go and pairing them with the car models.
We picked Xamarin Forms for this mobile application because it has near-native experience across multiple platforms.
Camera functionality was programmed separately for Android and iOS.
Android is using Tensorflow as an ML library.
iOS is using Vision which is native library developed by Apple.
The real value
Eight-fold increase in the quantity of searched vehicles with no demand of increased effort by the users.
Accessibility - Hakatron can be used by anybody with car and a smartphone.
No need to set up or manually input license plates. Just turn on the app and drive.
Services
Interface design
iOS and Android implementation
Testing in traffic
Releases to the Apple app store and Google Play
Tech support
Technology
iOS native ML core
Android native ML core
UI Xamarin Forms wrapper