Comparative analysis of background subtraction algorithms in the task of recognizing and identifying people in a video stream
I A Idimechev, А С Михалев, A. A. Goponenko, S S Stavenko, В В Кукарцев, В С Тынченко · Journal of Physics Conference Series · 2020
Abstract The article is devoted to one of the most urgent tasks for today: preprocessing video material in the process of recognizing and identifying people in a video stream. The authors of this article conducted a study, which consists in comparing the results of several algorithms for subtracting the background on various input video materials (from the office and the minimum concentration of people, to the dining room in which there are up to ten or more people). The data obtained during the analysis of the algorithms is compared (such data are the number of frames processed per second, and the average load of the computer’s central processor). The most efficient algorithms that coped with the task most quickly, accurately and with the least expenditure of resources, were identified, as well as those that coped with the task the worst.