To monitor violations of traffic rules will be the neural network

26 December 2018

Traffic violations in Moscow began to record using neural network

In many branches of humanity relies on artificial intelligence and neural networks. Now they began to record traffic violations in Moscow. It is expected that the network will help to avoid situations when the camera recognizes wrong license plate number, and calculate the fraud by the substitution of rooms. About high-tech innovation reports reported.

Director of photo and video fixation of the Moscow center of traffic organization Eugene Legler reported that the introduction of neural networks will allow to automatically verify the conformity mark and number of the car and eliminate the situation when the camera recognizes wrong license plate number.

In addition, the neural network will solve another big problem — the substitution of numbers for fraudulent activities.

First camera reads the vehicle's license plate, and it from the database, the traffic police was unloaded stamp machine. After that, the TMS staff were required to manually map the model of the car in the photo and from the database. Neural network automatiseret this process. To determine the brand takes into account the size of the vehicle and its shape.

According to the representative, that is, USING a neural network already trained to determine 103 the most popular regional brand. In total, they account for 95% of all reported car: from the familiar to the capital's roads "Germans" and "Japanese" to new machines Asian manufacturers.