Russian Neural Network that reads Dirty License Plates is being prepared to be tested on the M12 Highway and Central Ring Road

General Operator Ltd (part of Telematika Corporation) plans to pilot test the reading of dirty license plates by Russian software with elements of a neural network. Soft Telematika LLC (part of Telematika Corporation) developed the software for the Free Flow tolling system.

The test is ready to start at the beginning of next year on the Central Ring Road (CRR) and operating sections of the M12 highway. Software based on machine vision will read dirty license plates that could not be read at the tolling gantries. If the test outcomes are positive, General Operator will initiate corrections to the legal and regulatory framework to consider such data when rendering administrative rulings.

“Telematika’s existing technological and intellectual resources enable us to deploy innovative solutions. Software using a neural network will practically eliminate errors while reading dirty or deformed license plates. If necessary, it will be possible in the future to teach the technology to also detect intentionally concealed license plates on vehicles based on their distinctive features like the dimensions, scratches, nameplates, stickers, etc. To do so, the software will compare similar vehicles using the database of all vehicles that have ever travelled on the toll roads of Avtodor State Corporation,” shared Marsel Nigmetzyanov, General Director of General Operator Ltd.

In order to travel on the toll roads that use Free Flow, drivers do not need to stop at toll gates and waste time at tolling points. Vehicles pass through gantries equipped with cameras and sensors without slowing down. A toll ticket is automatically created according to the distance actually travelled on the basis of the vehicle classification and identified license plates. Since opening the first toll section to December 2022, Free Flow has processed more than 124.9 million transactions.