Review article

The paper considers an integrated approach toimproving road safety based on automated controlsystems. The tasks of detecting vehicles and faces ofroad traffic participants (drivers, pedestrians) as a keyelement for emergency prevention are analyzed. Theapplied computer vision algorithms (including YOLO),the methodology of integration with driver assistancesystems (ADAS), as well as the mechanisms forgenerating evidence-based media stream and analyticsfor subsequent analysis of accidents are described. Theresults of testing, the architecture of the software andhardware prototype and practical recommendations forthe implementation of the system are presented. Specialattention is paid to assessing the potential effectivenessof the solution for reducing accidents and taking intoaccount the real limitations during operation.