PNIPU: Program for instant reading of road signs and warning of drivers

Main article: Video analytics (terms, applications, technologies)

2024: Introduction of a program for instant reading of road signs

Scientists PNIPU have developed a program for instant reading of road signs and warning drivers. The university announced this on March 1, 2024.

For to data TRAFFIC POLICE the main reason - ROAD ACCIDENT violation of traffic rules by drivers. It accounts for almost 90% of the total number of accidents. Including violations occur due to the fact that drivers do not always see road signs in a timely manner. At the same time the Russian , manufacturers are still producing, cars which are not equipped with a system for recognizing them and issuing warnings to the driver about the road situation. And the purchase of a separate foreign device costs an average of 20 thousand. rubles PNIPU scientists have created a domestic system that recognizes road signs in real time and thereby reduces the risk of an accident. The device is designed to be installed in an already operated car, and over time can be built into the Russian production and assembly stage. transport

A certificate of state registration of the computer program No. 2024611982 by the Federal Service for Intellectual Property (Rospatent) was issued for development. The project was carried out as part of the Priority 2030 strategic academic leadership program.

Usually in cars, the road sign recognition system helps when the driver misses them, moving along any type of road. It tracks road signs using a digital camera located behind the rear view mirror data navigation and system. This is especially important for novice drivers who find it difficult to quickly read and respond to warnings when there is a large flow of cars on the roads.

Such systems are already being built into foreign business-class cars. Drivers of domestic cars can buy a separate gadget with such a function, but its cost varies from 20 thousand rubles.

PNIPU scientists have created their own algorithm for recognizing road signs using modern methods of computer vision and machine learning based on freely distributed software. The program is written in Python and weighs only 0.1 MB. The developers implemented software based on the OpenCV, Yolo and PyTorch libraries. The scientists then tested and optimized the system in a variety of lighting and pollution conditions.

As a result, we have a program that recognizes road signs in real time from a video stream. The visibility of the device depends on the quality characteristics of the lens used. With sufficient illumination, the signs are qualitatively read at a distance of low beam headlights, approximately 25 meters. The system then sends a signal voice assistant or projection onto the windshield. The driver has about 1-5 seconds to respond to the received. information Another plus of the device is that it has nothing to do Internet with or, which GPS means it works stably in any conditions, "explained Alexander Larionov, engineer of the Department of Information Technologies and Automated Systems of PNIPU.

As of March 2024, the project is at the stage of experimental research on vehicles. Then it is supposed to sell devices containing this software. The system will gradually "learn" the new signs using periodic updates, about the same as applications in the phone.

So PNIPU scientists have developed a program that increases road safety by reducing the number of violations associated with incorrect reading of road signs. Such a device can be mounted on any vehicle. In connection with automatic information about signs and markings, drivers will be able to save time and quickly respond to changes in the road situation. Thanks to the development of Perm polytechnics, the risk of road accidents will decrease, and the quality and safety of the domestic automotive industry will increase.