Nurseitov D., Bostanbekov K., Toiganbayeva N., Zhalgas A., Yedilkhan D., Amirgaliyev B.
Vision-Based People Counting and Tracking for Urban Environments // Journal of Imaging. — 2026. — Vol. 12, No. 1. — P. 27. — DOI: 10.3390/jimaging12010027.

Abstract: Population growth and expansion of urban areas increase the need for intelligent passenger traffic monitoring systems. Accurate estimation of the number of passengers is important for improving the efficiency, safety, and quality of transport services. This paper proposes an approach to automatic detection and counting of people using computer vision and deep learning methods. While YOLOv8 and DeepSORT have been widely explored individually, the contribution lies in a task-specific modification of the DeepSORT tracking pipeline optimized for dense passenger environments, occlusions, and dynamic lighting, as well as in a unified architecture integrating detection, tracking, and automatic event-log generation. A proprietary dataset of 4,047 images and 8,918 labeled objects achieved 92% detection accuracy and 85% counting accuracy. Compared to Mask R-CNN and DETR, YOLOv8 demonstrates an optimal balance between speed, accuracy, and computational efficiency. The results confirm that computer vision can effectively replace traditional sensor-based passenger counting systems. Future work includes expanding the dataset, introducing multicamera integration, and adapting the model for embedded devices.

Link / DOI: https://doi.org/10.3390/jimaging12010027