Abstract
The article discusses current R&D on the intellectualization of traffic lights equipped with IP video cameras. It shows the application areas of smart traffic lights and describes the stages of designing and semi-industrial testing of the prototype, the core of which is a Raspberry Pi 4 mini PC. During the design and construction of the hardware and software complex (HSC), features of frame-by-frame methods for webcams and streaming methods for IP video cameras for AI image recognition were identified. To increase vandal, water, and dust resistance and reliability for semi-industrial testing of the prototype, an outdoor IP video camera by HIKVISION in the EXIR Fixed Mini Bullet Network Camera modification was used. Periodic power outages at intersections in Almaty required research and development of a method for automatic start-up of the HSC from the moment of power loss. The proposed BIOS configuration algorithm for the Raspberry Pi 4 mini PC is novel and unique in its OS Raspbian settings as a non-standard version of Debian GNU/Linux. The found solutions and test results demonstrated the reliability of the automatic start-up of the traffic light with neural computer vision. Smart traffic lights with video vision allow adaptive control of traffic lights with AI in street conditions at the Abay/Saina interchange in the metropolis of Almaty.