Plate number recognition

Link for video demo.

Analysis of traffic on the streets

Detection and classification of vehicles

Detection and recognition of plate numbers

Support of regional standards of plate number naming

Optimized engine for aggregation, storing and monitoring of data from 1000+ cameras.

Realtime manual monitoring of traffic from the streets requires huge amount of time and effort.

We trained neural networks in Tensorflow which detect vehicles, plate numbers and letters with digits. Our models are integrated in an optimized multithreaded C++ agent which communicates with the server via grpc protocol. We have also implemented a cross-platform (windows/mac/linux) client for data aggregation, storing and monitoring information from different cameras