An AI-Based Smart CCTV System for Classroom Energy Monitoring and Management

Abstract

Electricity wastage in the educational institutions is a problem that is often ignored, wherein ceiling fans are often not switched off even after classrooms are vacated. Manual monitoring of such situations is not efficient, and can be a challenge to keep up over a large campus, leading to unnecessary energy consumption. Although a number of occupancy detection systems based on computer vision have been proposed, most of them only deal with the recognition of the human presence; there is little dealing with the verification of the true operating state of electrical appliances. To overcome this drawback, this paper introduces an AI-based smart classroom monitoring mechanism based on the existing CVC system infrastructure, capable of detecting the human presence and ceiling fan operation and report them, in real time. The proposed system is constructed according to a modified YOLOv8 deep learning system with a custom classroom dataset for a two-object detection. In addition, a simple motion analysis technique is used within the region detected fan to identify if the fan is running or stationary. By combining the detection of people with the verification of the state of the appliances, the system added a context-aware decision mechanism that generates alerts only when a fan is found to be operating in an empty classroom. Such an approach helps in reducing false alarms while also focused on the direct root cause of wastage of electricity. The proposed solution is cost-effective, scalable and can be deployed in smart classrooms and academic campuses to aid efficient energy management.

Introduction

The rapid expansion of educational institutions has significantly increased electricity consumption, particularly in classrooms equipped with ceiling fans and other appliances. A common issue is unnecessary energy wastage, as fans often remain switched on even after classrooms are vacated. While leaving one fan running may seem minor, repeated occurrences across multiple classrooms lead to substantial financial and environmental consequences. Traditional energy management methods rely heavily on manual supervision or basic occupancy detection systems, which either require constant monitoring or fail to verify whether appliances are actually operating.

To address these limitations, the paper proposes Smart Class Sense, an AI-based smart classroom monitoring system that uses existing CCTV infrastructure for context-aware energy management. Unlike traditional systems that detect only human presence, this system integrates both human detection and appliance (ceiling fan) status verification. Using computer vision techniques, OpenCV, and a YOLOv8-based deep learning model, the system first detects whether people are present. If no one is detected, it checks whether the ceiling fan is running through a custom-trained fan detection model and motion analysis. Alerts are generated only when a fan is operating in an empty classroom.

The system architecture is modular, scalable, and cost-effective since it utilizes existing CCTV cameras. Its decision logic ensures minimal false alarms by considering both occupancy and appliance status. Experimental results demonstrate reliable human detection, accurate fan recognition, and effective identification of electricity wastage scenarios.

Key advantages include low deployment cost, real-time monitoring, reduced human involvement, high reliability, and context-aware decision-making. The system is suitable for schools, colleges, universities, and other indoor environments such as libraries and laboratories. Overall, Smart Class Sense provides a practical, intelligent, and efficient solution for reducing electricity wastage in educational institutions.

Conclusion

Electricity wastage in classroom is one of the most common and serious issues in educational institutions. Manual classroom monitoring is not reliable and tends to be overlooked because of time and manpower limitations. In this study, the concept of Smart class sense is introduced as an intelligent and automatic solution for energy savings on electricity consumption in classrooms. The proposed system manages to merge human presence detection and ceiling fan operation verification by computer vision and deep learning techniques. Using existing CCTV cameras is another advantage in that the system does not require extra hardware cost, and is still simple to deploy. The context aware decision logic makes sure that the alerts are pushed out only when the fan is running in an empty classroom, resulting in less false alarms and better reliability. The results and observations show that the system works well in the real classroom-like conditions. Even with simple logic and light models, the system is able to spot electricity wastage scenarios correctly. Overall, Smart Class Sense is an effective, low cost, and practical solution for smart classroom energy monitoring, particularly for large campuses to which manual supervision is difficult.

References

[1] Beibut Amirgaliyev, Miras Mussabek, Tomiris Rakhimzhanova, and Ainur Zhumadillayeva. A review of machine learning and deep learning methods for person detection, tracking and identification, and face recognition with applications. Sensors, 25(5):1410, 2025. [2] [W. Zhang, J. Calautit, P. W. Tien, Y. Wu, and S. Wei, “Deep learning models for vision-based occupancy detection in high-occupancy buildings,” Journal of Building Engineering, vol. 98, Art. no. 111355, 2024. [3] C. E. Kim, M. M. Dar Oghaz, J. Fajtl, V. Argyriou, and P. Remagnino, “A comparison of embedded deep learning methods for person detection,” arXiv preprint arXiv:1812.03451, 2019. [4] O. Zeleny, T. Fryza, T. Bravenec, and G. Nair, “Detection of room occupancy in smart buildings,” Radioengineering, vol. 33, no. 3, pp. 432–441, 2024. [5] M. Safaldin, N. Zaghdhen, and M. Mejdoub, “An improved YOLOv8 to detect moving objects,” IEEE Access, vol. 12, 2024, doi: 10.1109/ACCESS.2024.3393835. [6] T. A. R. Shyaa and A. A. Hashim, “Enhancing real human detection and people counting using YOLOv8,” BIO Web of Conferences, vol. 97, Art. no. 00061, 2024. [7] T. Chavdarova and F. Fleuret, “Deep multi-camera people detection,” arXiv preprint arXiv:1702.04593, 2017. [8] R. Han, J. Zhao, W. Feng, Y. Gan, L. Wan, and S. Wang, “Complementary-view co-interest person detection,” in Proc. 28th ACM Int. Conf. on Multimedia (MM ’20), 2020, pp. 1–9. [9] Y.-L. Hou and G. K. H. Pang, “People counting and human detection in a challenging situation,” IEEE Trans. Systems, Man, and Cybernetics—Part A, vol. 41, no. 1, pp. 24–33, 2011. [10] X. Cui, V. G. Zakka, and M. Lee, “A computer vision-based model for occupancy detection using low-resolution thermal images,” in Proc. Int. Conf. on Building Energy and Environment, 2025. [11] M. Rodriguez, I. Laptev, J. Sivic, and J.-Y. Audibert, “Density-aware person detection and tracking in crowds,” in Proc. IEEE Int. Conf. on Computer Vision (ICCV), 2011. [12] A. Sarkar and P. Agarwal, “Real-time object detection using YOLOv8 with webcam integration,” International Journal of Research Publication and Reviews, vol. 6, no. 5, pp. 1165–1168, 2025. [13] A. Yao, J. Gall, C. Leistner, and L. Van Gool, “Interactive object detection,” in Proc. IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), 2012, pp. 3242–3249. [14] W. R. Schwartz, A. Kembhavi, D. Harwood, and L. S. Davis, “Human detection using partial least squares analysis,” in Proc. IEEE Int. Conf. on Computer Vision (ICCV), 2009. [15] M. Andriluka, S. Roth, and B. Schiele, “People-tracking-by-detection and people-detection-by-tracking,” in Proc. IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), 2008. [16] I. S. Topkaya, H. Erdogan, and F. Porikli, “Counting people by clustering person detector outputs,” in Proc. IEEE Int. Conf. on Advanced Video and Signal Based Surveillance (AVSS), 2014, pp. 313–318. [17] J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, “You only look once: Unified, real-time object detection,” in Proc. IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), 2016, pp. 779–788. [18] M. D. Solbach and J. K. Tsotsos, “Vision-based fallen person detection for the elderly,” in Proc. IEEE Int. Conf. on Computer Vision (ICCV), 2017, pp. 1433–1441. [19] P. A. K. Ansah et al., “SB-YOLO-V8: A multilayered deep learning approach for real-time human detection,” Engineering Reports, vol. 7, no. 2, Art. no. e70033, 2025. [20] K. Cui et al., “Research on intelligent recognition algorithm of college students’ classroom behavior based on improved single shot multibox detector,” IEEE Access, vol. 12, pp. 168106–168120, 2024. [21] F. Pasti and N. Bellotto, “Evaluation of computer vision-based person detection on low-cost embedded systems,” in Proc. Int. Conf. on Computer Vision Theory and Applications (VISAPP), 2023. [22] Chloe Eunhyang Kim, Mahdi Maktab Dar Oghaz, Jiri Fajtl, Vasileios Argyriou, and Paolo Remagnino. A comparison of embedded deep learning methods for person detection. arXiv preprint arXiv:1812.03451, 2018 [23] M. Talib, A. H. Y. Al-Noori, and J. Suad, “YOLOv8-CAB: Improved YOLOv8 for real-time object detection,” Karbala International Journal of Modern Science, vol. 10, no. 1, Art. no. 5, 2024. [24] Alejandro S Mart´?nez-Sala, Lucio Hernando-Canovas, Juan C S´ anchez-´ Aarnoutse, and Juan J Alcaraz-Esp´?n. Resource-efficient fog computing vision system for occupancy monitoring: A real-world deployment in university libraries. Internet of Things, page 101748, 2025. [25] Joseph Redmon, Santosh Divvala, Ross Girshick, and Ali Farhadi. You only look once: Unified, real-time object detection. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 779–788, 2016 [26] Mukaram Safaldin, Nizar Zaghden, and Mahmoud Mejdoub. An improved yolov8 to detect moving objects. IEEE Access, 12:59782– 59806, 2024.Mikel Rodriguez, Ivan Laptev, Josef Sivic, and Jean-Yves Audibert. [27] Francesco Pasti, Nicola Bellotto, Evaluation of computer vision-based person detection on low-cost embedded systems,Proceedings of the International Conference on Pattern Recognition Applications and Methods (ICPRAM), 2023. [28] Tahreer Abdul Ridha Shyaa and Ahmed A Hashim. Enhancing real human detection and people counting using yolov8. In BIO Web of Conferences, volume 97, page 00061. EDP Sciences, 2024. [29] H. H. Nguyen, V. T. Bui, T. N. Ta, H. M. Pham, N. C. Nguyen, and D.M.Nguyen,“YOLO based real-time human detection for smart video surveillance at the edge,”in Proc. IEEE Int. Conf. on Consumer Electronics (ICCE), 2020. [30] R. Han, J. Zhao, W. Feng, Y. Gan, L. Wan, and S. Wang,“Complementary-view co-interest person detection,”in Proceedings of the 28th ACM International Conference on Multimedia (MM ’20),Seattle, WA, USA, Oct. 2020, pp. 1–9, doi: 10.1145/3394171.3413659.

Copyright

Copyright © 2026 Pushpendra Mishra, Aryan Ahirwar, Anshul Kumbhare, Nayan Tandekar, Piyush Rahangdale, Sumit Sonar. 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.