Real-Time Visitor Movement Verification Using Floor-Level Camera Detection

Abstract

Visitor monitoring plays a crucial role in maintaining security across modern infrastructures such as offices, educational institutions, hospitals, and public facilities. Traditional surveillance systems primarily record video footage and rely on continuous manual monitoring, rendering them inefficient and susceptible to human error. Moreover, such systems lack the capability to automatically verify visitor movement across different areas or generate real-time alerts for policy violations. This paper proposes a Real-Time Visitor Movement Verification System (RVMVS) using floor-level camera detection. The system employs mobile cameras positioned at floor level to capture live video streams from multiple locations simultaneously. Captured frames are processed using advanced computer vision techniques to detect human presence in real time, leveraging YOLOv8 for accurate person detection and OpenCV for video frame processing and analysis. Upon detection, the system automatically labels the visitor, captures an image, and records visitor information alongside a precise timestamp. A multi-camera tracking mechanism verifies visitor movement across camera zones to ensure accurate and uninterrupted monitoring. The system further incorporates threshold time monitoring, automated alert generation, and comprehensive log management. Experimental evaluation demonstrates a detection accuracy of 91%, tracking accuracy of 88%, and an alert response time of 2–3 seconds at 24 frames per second. This intelligent monitoring approach substantially reduces dependence on manual surveillance, improves operational security efficiency, and provides a scalable solution for real-world deployment.

Introduction

The text presents a Real-Time Visitor Movement Verification System (RVMVS) designed to modernize surveillance and visitor management in large facilities using computer vision and deep learning.

Traditional CCTV systems depend heavily on human monitoring, which is inefficient, error-prone, and not capable of real-time identity tracking. To overcome this, the proposed system uses YOLOv8 for person detection, DeepSORT for tracking, and Re-ID features for cross-camera identity matching, enabling automated monitoring of visitor movement across multiple zones.

The system workflow includes visitor registration, real-time detection, multi-camera tracking, threshold-based time monitoring, alert generation, and automatic log creation. Mobile phones are used as low-cost IP cameras, while a central GPU-powered server processes video streams. A web dashboard provides real-time visualization, alerts, and visitor statistics.

During evaluation with simulated scenarios, the system achieved around 91% detection accuracy and 88% tracking accuracy, processing video at 24 FPS with low latency (<100 ms). It successfully tracked multiple visitors across zones and generated alerts for unauthorized movement and overstays.

Compared to traditional CCTV and manual monitoring, RVMVS offers fully automated detection, tracking, alerting, and logging, with higher scalability and lower human error. However, performance decreases under low light and heavy occlusion, and identity matching may require further optimization for specific environments.

Conclusion

This paper presented the Real-Time Visitor Movement Verification System (RVMVS), a comprehensive, cost-effective intelligent surveillance platform designed for real-world deployment in security-sensitive institutional environments. The system integrates YOLOv8-based person detection, multi-camera Re-ID tracking, automated visitor registration, threshold time monitoring, structured alert management, and detailed log generation into a unified, modular architecture that operates over standard Wi-Fi infrastructure using commercially available mobile cameras. Experimental evaluation over 147 simulated visitor sessions demonstrated that the system achieves a person detection accuracy of 91%, a cross-camera tracking accuracy of 88%, a processing throughput of 24 FPS, and an end-to-end alert latency of under 100 milliseconds. Comparative analysis confirmed that the RVMVS outperforms conventional CCTV and manual monitoring approaches across all evaluated dimensions, including automation level, scalability, and error resistance. The system\'s user-friendly web dashboard and role-based access control further enhance operational usability and institutional security governance. The proposed system makes several contributions to the field of intelligent surveillance. First, it demonstrates the practical viability of deploying YOLOv8-based detection in a multi-camera, real-time visitor management context using consumer-grade hardware. Second, it introduces a structured, modular architecture that cleanly separates detection, tracking, registration, alerting, and logging concerns, facilitating future upgrades and extensions. Third, it provides quantitative benchmarking of system performance across multiple operational scenarios, providing a baseline for comparison with future work. Certain limitations identified during evaluation indicate important directions for future research. Enhancing detection robustness under low-light conditions through night-vision camera integration or image enhancement preprocessing represents a near-term priority. Improving cross-camera tracking accuracy in high-density scenarios may be addressed through the adoption of transformer-based Re-ID architectures or graph neural network-based association algorithms. Integration of edge computing capabilities—deploying lightweight detection models directly on camera nodes—would reduce network bandwidth requirements and improve system scalability for large facilities. Longer-term research directions include the integration of behavioral analytics to detect anomalous visitor movement patterns, incorporation of natural language processing for automated incident report generation from alert logs, and exploration of federated learning approaches to enable privacy-preserving model improvement across multiple deployment sites. Cloud-based deployment architectures would further enhance system scalability and enable centralized monitoring of geographically distributed facilities. The RVMVS represents a meaningful step toward fully automated, intelligent, and accountable visitor surveillance, and its open modular design provides a strong foundation for continued advancement in this important application domain.

References

[1] J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, “You Only Look Once: Unified, Real-Time Object Detection,” in Proc. IEEE Conf. Computer Vision and Pattern Recognition (CVPR), 2016, pp. 779–788. [2] “Real-Time Object Detection using YOLO,” International Journal of Computer Applications, 2021. [3] “Object Detection and Tracking using Deep Learning and OpenCV in Real-Time Environment,” International Journal of Engineering Research & Technology (IJERT), vol. 12, no. 4, 2023. [4] G. Jocher, A. Chaurasia, and J. Qiu, “Ultralytics YOLOv8,” GitHub repository, 2023. [Online]. Available: https://github.com/ultralytics/ultralytics [5] N. Wojke, A. Bewley, and D. Paulus, “Simple Online and Realtime Tracking with a Deep Association Metric,” in Proc. IEEE Int. Conf. Image Processing (ICIP), 2017, pp. 3645–3649. [6] “A Novel Approach on Object Tracking for Face Recognition,” International Journal of Engineering Research & Technology (IJERT), vol. 8, no. 6, 2019. [7] “Enhancing Attendance Management Systems using Facial Recognition,” International Journal of Engineering Research & Technology (IJERT), vol. 13, no. 2, 2024. [8] IJEDR1503008, “Object Tracking System,” International Journal of Engineering Development and Research, vol. 3, no. 2, 2015. [9] IJRESM_V7_I2_4, “Deep Learning-Based Surveillance Systems: A Review,” International Journal of Research in Engineering, Science and Management, vol. 7, no. 2, 2024. [10] IJNRD2407062, “Multi-Camera Person Re-Identification using Appearance Features,” International Journal of Novel Research and Development, vol. 9, no. 7, 2024. [11] “AI-Based Surveillance System using Computer Vision Techniques,” International Journal of Advanced Research in Computer Science, vol. 13, no. 3, 2022. [12] “Multi-Camera Object Tracking in Surveillance Systems,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 31, no. 8, pp. 3124–3138, 2021. [13] “Smart Surveillance System using Deep Learning Techniques,” International Journal of Innovative Technology and Exploring Engineering (IJITEE), vol. 11, no. 5, 2022. [14] “Visitor Monitoring System using Computer Vision,” International Journal of Computer Science and Mobile Computing, vol. 12, no. 3, 2023. [15] “Real-Time Human Detection and Tracking in Video Surveillance Systems,” Journal of Visual Communication and Image Representation, vol. 82, 2022. [16] “Automated Visitor Management System using Computer Vision,” International Journal of Engineering and Technology, vol. 10, no. 1, 2023. [17] “Multi-Camera Tracking for Real-Time Surveillance Applications,” Pattern Recognition Letters, vol. 153, pp. 112–119, 2022. [18] “Intelligent Alert Management in Smart Surveillance Systems,” IEEE Access, vol. 11, pp. 24512–24523, 2023. [19] “Performance Analysis of Deep Learning-Based Object Detection Systems for Surveillance,” Computers & Electrical Engineering, vol. 98, 2022. [20] T. C. Hui, “Multi-Camera Tracking and Surveillance System Design and Implementation,” M.Eng. thesis, Dept. Electrical Eng., Multimedia University, Malaysia, 2023.

Copyright

Copyright © 2026 Mr. Anway K. Yerawar, Ms. Suhani S. Jaiswal, Ms. Sonali R. Jangid, Mr. Hardik S. Agrawal, Dr. A. P. Bakshi. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.