MTUSI: Neural network for recognizing car brands

Main article: Neural networks (neural networks)

2023: Neural Network Development

Employees of the Faculty of Information Technologies of MTUSI, under the guidance of Faculty Dean Mikhail Gorodnichev, have developed a neural network created specifically to solve the problem of recognizing vehicle brands. The university announced this on January 31, 2023.

Human life is an extensive list of practical tasks that can be automated to improve overall efficiency. In the past, this list was limited only to tasks, the solution of which did not require creative thinking and was characteristic of only one person. At this stage, scientific and technological progress over the past two decades has greatly expanded this list.

Especially for such tasks, MTUSI employees created convolutional neural networks, or CNN (from the English Convolutional neural network). Their task is to receive images as input information and, based on the results of their work, give names of classes of objects that were previously determined in the learning process using the robas loss function. In the process of developing a neural network, data was collected from the Auto.ru service and outdoor video surveillance cameras, and DataSet itself was collected with a size of more than 90 thousand copies, which were subsequently placed and pre-processed, thanks to which the developed technology is able to determine cars and their brands by individual elements to improve accuracy.

Artificial intelligence for January 2023 is one of the most promising areas in the IT region. One of the advantages of this development is its accuracy. In the future, the neural network is easily able to facilitate the processing of the incoming video stream for deeper collection of information about the composition of the transport stream, which will allow it to be controlled more optimally and safely.