Instruments and Systems: Monitoring, Control, and Diagnostics Annotation << Back Basic Processes of the YOLOv5 Algorithm in Video Surveillance Systems A.U. MENTSIEV, T.G. AYGUMOV, SH.A. ABDURASHIDOV Realizing accurate people counting in crowded places has a wide range of application capabilities, such as crowd management, security and resource optimization. The current research paper presents a comprehensive method for developing a real-time people counting system using YOLOv5 (You Only Look Once) object detection technology. In this system, convolutional neural networks (CNNs) are used to detect people in CCTV footage, providing a reliable mechanism for counting people in real-world scenarios. The study includes a detailed outline of the steps such as data preparation, model training, integration, deployment and continuous monitoring required to formulate an effective and accurate people counting method. Keywords: artifi cial intelligence, YOLOv5 algorithm, object detection, machine learning. DOI: 10.25791/pribor.12.2023.1459 Pp. 15-18.