Research Article
Open Access
An Enhanced YOLOv9-Based Detection Method and Warning System for Indoor Electric Motorcycles
1
College of Electronic Science and Technology, National University of Defense Technology, Changsha 410073, China
2
Key Laboratory of Automatic Target Recognition, National University of Defense Technology, Changsha 410073, China
* Corresponding Author:
Moufa Hu,
[email protected]
This article belongs to the Special Topic:
Pattern Recognition and Information Fusion
Volume 3, Issue 2
2026
Received: 15 January 2026
Accepted: 08 June 2026
Published: 22 June 2026
Article Information
Published in
Chinese Journal of Information Fusion
Volume/Issue
Volume 3, Issue 2, 2026
Pages
153-165
Abstract
To address severe fire safety risks caused by electric motorcycles (EMs) and their batteries being illegally brought into building elevators, this paper presents a real-time EM detection and alarm system for elevator environments, built upon a multi-source information fusion framework and an improved YOLOv9. To elevate detection accuracy for EMs in confined elevator spaces, two core optimizations are embedded into the network: the Programmable Gradient Information (PGI) training strategy, and a lightweight Generalized Efficient Layer Aggregation Network (GELAN) backbone enhanced with depthwise separable convolution (DSConv). A dedicated dataset consisting of roughly 2,000 images is established for model training and validation. This dataset covers a wide range of elevator scenes, diverse target postures, and common occlusion cases. Experimental results show that the proposed model achieves [email protected] values of 0.952, 0.915 and 0.887 across three challenging elevator scenarios. The average per-frame inference latency is less than 35 ms, which fully meets the real-time constraints for edge device deployment. To construct a complete closed-loop detection and alarm system, we further design an alarm subsystem empowered by multi-source information fusion. It adopts dual-threshold decision logic and supports multimodal alerts including audio, visual cues, and log recording. This framework first judges the presence of EMs via a basic confidence threshold, then verifies the temporal continuity of detection results using a dedicated alarm threshold, and finally triggers multimodal alarms accordingly. This system effectively improves the efficiency of alarm notifications and provides a practical technical solution for fire safety management of elevators in residential and commercial buildings.
Graphical Abstract
Keywords
electric motorcycles detection
YOLOv9
real-time system
intelligent alarm
multi-source information fusion
Data Availability Statement
Data will be made available on request.
Funding
This work was supported without any funding.
Conflicts of Interest
The authors declare no conflicts of interest.
AI Use Statement
DeepSeek-R1 was used for proofreading the entire manuscript and for checking the formatting of references. DeepL Translator was used for translating references originally published in Chinese into English. No generative AI was used for content generation, experimental design, data analysis, or interpretation of results. The authors take full responsibility for the content, accuracy, and originality of this manuscript.
Ethical Approval and Consent to Participate
The elevator surveillance footage used in this study was obtained with formal authorization from the property management companies of the participating residential buildings, in accordance with their existing security camera operation policies, which include standard notice
to residents that footage may be used for safety research and system improvement purposes. The research team did not independently install any recording equipment; all data were extracted from pre-existing, authorized building surveillance systems. No additional personal information beyond visual imagery was collected, and the study does not involve any direct interaction with or recruitment of human participants. Individuals incidentally captured in the published figures have been informed through standard building notices that surveillance recordings may be used for research purposes; where feasible, facial regions have been obscured in the published figures to further protect individual privacy.
References
- Sheng-ping, T., Xu-hong, H., Nan, Z., Wei-hao, X., Yan-yi, C., & Cun-xin, Z. H. U. (2022, November). Research on two-wheeled vehicle entry ban system based on deep learning. In International Conference on Advanced Intelligent Systems and Informatics (pp. 224-233). Cham: Springer International Publishing.
[CrossRef] [Google Scholar] - Li, L., Liu, B., Zheng, W., Wu, X., Song, L., & Dong, W. (2022). Investigation and numerical reconstruction of a full-scale electric bicycle fire experiment in high-rise residential building. Case Studies in Thermal Engineering, 37, 102304.
[CrossRef] [Google Scholar] - Wang, Y., Zhang, Y., & Yu, Z. (2022, August). Fire risk assessment of electric bicycles charging facilities in old urban communities. In 2022 Power System and Green Energy Conference (PSGEC) (pp. 447-451). IEEE.
[CrossRef] [Google Scholar] - Han, Y., & Wang, W. (2025). Identification System for Electric Bicycle in Compartment Elevators. Electronics, 14(13), 2638.
[CrossRef] [Google Scholar] - Li, T., Lei, L., Wang, Z., Shi, P., & Wu, Z. (2025). An efficient improved YOLOv10 algorithm for detecting electric bikes in elevators. Electronic Research Archive, 33(6), 3673-3698.
[CrossRef] [Google Scholar] - Huang, J., Lin, G., Dong, Z., & Sun, F. (2025, June). Research on Real-time Object Detection Algorithms for Elevator Intelligent Monitoring Scenarios. In Proceedings of the 2025 2nd International Conference on Computer and Multimedia Technology (pp. 229-235).
[CrossRef] [Google Scholar] - Redmon, J., Divvala, S., Girshick, R., & Farhadi, A. (2016, June). You Only Look Once: Unified, Real-Time Object Detection. In 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (pp. 779-788). IEEE Computer Society.
[CrossRef] [Google Scholar] - Wang, C. Y., Bochkovskiy, A., & Liao, H. Y. M. (2023, June). YOLOv7: Trainable Bag-of-Freebies Sets New State-of-the-Art for Real-Time Object Detectors. In 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (pp. 7464-7475). IEEE.
[CrossRef] [Google Scholar] - Zhang, X., Yuan, D., Sun, H., Yu, L., Guo, W., & Feng, X. (2025, June). YOLOv5-Based Detection of Electric Vehicles and Batteries in Residential Elevators for Fire Hazard Prevention. In 2025 5th International Conference on Artificial Intelligence, Big Data and Algorithms (CAIBDA) (pp. 1-8). IEEE.
[CrossRef] [Google Scholar] - He, Y. (2024). Automatic detection of electric motorcycle based on improved YOLOv5s network. Journal of Electrical and Computer Engineering, 2024(1), 4889707.
[CrossRef] [Google Scholar] - Saleh, K., Szénási, S., & Vámossy, Z. (2021, January). Occlusion handling in generic object detection: A review. In 2021 IEEE 19th World Symposium on Applied Machine Intelligence and Informatics (SAMI) (pp. 000477-000484). IEEE.
[CrossRef] [Google Scholar] - Zhu, Z., Cao, J., Hao, T., Zhai, W., Sun, B., Jia, G., & Li, M. (2020). Highly secure edge-intelligent electric motorcycle management system for elevators. Journal of Cloud Computing, 9(1), 41.
[CrossRef] [Google Scholar] - Liu, P., Li, Y., Yuan, L., & Wang, Y. (2026). A Lightweight Method for Detecting the Behavior of Electric Bicycle Entering Elevators Based on Deformation Multi-scale Collaboration. In International Conference on Computing and Pattern Recognition (pp. 393-403). Springer, Singapore.
[CrossRef] [Google Scholar] - Lin, Y., Chen, X., Zhong, W., & Pan, Z. (2021, March). Online detection system for electric bike in elevator or corridors based on multi-scale fusion. In 2021 4th International Conference on Advanced Electronic Materials, Computers and Software Engineering (AEMCSE) (pp. 31-34). IEEE.
[CrossRef] [Google Scholar] - Wan, Y., Hu, Y., Li, X., Song, Z., & Hou, T. (2024, November). Improved Algorithm of Elevator Blocking System Based on YOLOv8. In 2024 12th International Conference on Information Systems and Computing Technology (ISCTech) (pp. 1-5). IEEE.
[CrossRef] [Google Scholar] - Zhao, Z., Li, S., Wu, C., & Wei, X. (2023). Research on the rapid recognition method of electric bicycles in elevators based on machine vision. Sustainability, 15(18), 13550.
[CrossRef] [Google Scholar] - Zhu, K., Dai, J., & Gu, Z. (2024). Dynamic tracking method based on improved deepsort for electric vehicle. World Electric Vehicle Journal, 15(8), 374.
[CrossRef] [Google Scholar] - Wang, C. Y., Yeh, I. H., & Mark Liao, H. Y. (2024, September). Yolov9: Learning what you want to learn using programmable gradient information. In European conference on computer vision (pp. 1-21). Cham: Springer Nature Switzerland.
[CrossRef] [Google Scholar] - Ikmel, G., & El Amrani, E. L. (2024, June). Performance analysis of YOLOv5, YOLOv7, YOLOv8, and YOLOv9 on road environment object detection: comparative study. In 2024 international conference on ubiquitous networking (UNet) (Vol. 10, pp. 1-5). IEEE.
[CrossRef] [Google Scholar] - Wang, L., Letchmunan, S., & Xiao, R. (2024). Gelan-SE: squeeze and stimulus attention based target detection network for gelan architecture. IEEE Access, 12, 182259-182273.
[CrossRef] [Google Scholar] - Chen, Z., Fan, K., Ye, J., Xu, Z., & Wei, Y. (2025). A Lightweight Multi-Module Collaborative Optimization Framework for Detecting Small Unmanned Aerial Vehicles in Anti-Unmanned Aerial Vehicle Systems. Drones, 10(1), 20.
[CrossRef] [Google Scholar] - Li, Y., Zheng, Q., Xu, S., Wang, P., Wang, Y., Song, Z., & Guo, M. (2025). YOLO-EDGE: an object detection algorithm for traffic scenarios. The Journal of Supercomputing, 81(7), 840.
[CrossRef] [Google Scholar] - Liu, S., Qi, L., Qin, H., Shi, J., & Jia, J. (2018, June). Path Aggregation Network for Instance Segmentation. In 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 8759-8768). IEEE.
[CrossRef] [Google Scholar] - KIM, T., Sun-Been, P., Yu-Jeong, J., & Da-Eun, L. (2024). A study on the elevator system using real-time object detection technology YOLOv5. Journal of Internet of Things and Convergence, 2024, 10(2), 103–108.
[CrossRef] [Google Scholar] - Yu, X., Chen, C., Cheng, S., & Li, J. (2025). A lightweight detection method for X-ray security inspection based on YOLOV8. Applied Soft Computing, 113822.
[CrossRef] [Google Scholar] - Su, J., Yang, M., & Tang, X. (2024). Integration of ShuffleNet V2 and YOLOv5s networks for a lightweight object detection model of electric bikes within elevators. Electronics, 13(2), 394.
[CrossRef] [Google Scholar] - Long, X., Hu, S., Hu, Y., Gu, Q., & Ishii, I. (2019). An FPGA-based ultra-high-speed object detection algorithm with multi-frame information fusion. Sensors, 19(17), 3707.
[CrossRef] [Google Scholar] - Torralba, A., Russell, B. C., & Yuen, J. (2010). Labelme: Online image annotation and applications. Proceedings of the IEEE, 98(8), 1467-1484.
[CrossRef] [Google Scholar] - Bochkovskiy, A., Wang, C. Y., & Liao, H. Y. M. (2020). Yolov4: Optimal speed and accuracy of object detection. arXiv preprint arXiv:2004.10934.
[CrossRef] [Google Scholar] - Szegedy, C., Vanhoucke, V., Ioffe, S., Shlens, J., & Wojna, Z. (2016, June). Rethinking the Inception Architecture for Computer Vision. In 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (pp. 2818-2826). IEEE.
[CrossRef] [Google Scholar] - Loshchilov, I., & Hutter, F. (2017). Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101.
[CrossRef] [Google Scholar]
Cite This Article
APA Style
Luo, Q., Lu, P., Sun, Y., Yin, Z., & Hu, M. (2026). An Enhanced YOLOv9-Based Detection Method and Warning System for Indoor Electric Motorcycles. Chinese Journal of Information Fusion, 3(2), 153-165. https://doi.org/10.62762/CJIF.2026.420972
Export Citation
RIS Format
Compatible with EndNote, Zotero, Mendeley, and other reference managers
TY - JOUR AU - Luo, Qin AU - Lu, Pengxiang AU - Sun, Yibo AU - Yin, Zhichao AU - Hu, Moufa PY - 2026 DA - 2026/06/22 TI - An Enhanced YOLOv9-Based Detection Method and Warning System for Indoor Electric Motorcycles JO - Chinese Journal of Information Fusion T2 - Chinese Journal of Information Fusion JF - Chinese Journal of Information Fusion VL - 3 IS - 2 SP - 153 EP - 165 DO - 10.62762/CJIF.2026.420972 UR - https://www.icck.org/article/abs/CJIF.2026.420972 KW - electric motorcycles detection KW - YOLOv9 KW - real-time system KW - intelligent alarm KW - multi-source information fusion AB - To address severe fire safety risks caused by electric motorcycles (EMs) and their batteries being illegally brought into building elevators, this paper presents a real-time EM detection and alarm system for elevator environments, built upon a multi-source information fusion framework and an improved YOLOv9. To elevate detection accuracy for EMs in confined elevator spaces, two core optimizations are embedded into the network: the Programmable Gradient Information (PGI) training strategy, and a lightweight Generalized Efficient Layer Aggregation Network (GELAN) backbone enhanced with depthwise separable convolution (DSConv). A dedicated dataset consisting of roughly 2,000 images is established for model training and validation. This dataset covers a wide range of elevator scenes, diverse target postures, and common occlusion cases. Experimental results show that the proposed model achieves [email protected] values of 0.952, 0.915 and 0.887 across three challenging elevator scenarios. The average per-frame inference latency is less than 35 ms, which fully meets the real-time constraints for edge device deployment. To construct a complete closed-loop detection and alarm system, we further design an alarm subsystem empowered by multi-source information fusion. It adopts dual-threshold decision logic and supports multimodal alerts including audio, visual cues, and log recording. This framework first judges the presence of EMs via a basic confidence threshold, then verifies the temporal continuity of detection results using a dedicated alarm threshold, and finally triggers multimodal alarms accordingly. This system effectively improves the efficiency of alarm notifications and provides a practical technical solution for fire safety management of elevators in residential and commercial buildings. SN - 2998-3371 PB - Institute of Central Computation and Knowledge LA - English ER -
BibTeX Format
Compatible with LaTeX, BibTeX, and other reference managers
@article{Luo2026An,
author = {Qin Luo and Pengxiang Lu and Yibo Sun and Zhichao Yin and Moufa Hu},
title = {An Enhanced YOLOv9-Based Detection Method and Warning System for Indoor Electric Motorcycles},
journal = {Chinese Journal of Information Fusion},
year = {2026},
volume = {3},
number = {2},
pages = {153-165},
doi = {10.62762/CJIF.2026.420972},
url = {https://www.icck.org/article/abs/CJIF.2026.420972},
abstract = {To address severe fire safety risks caused by electric motorcycles (EMs) and their batteries being illegally brought into building elevators, this paper presents a real-time EM detection and alarm system for elevator environments, built upon a multi-source information fusion framework and an improved YOLOv9. To elevate detection accuracy for EMs in confined elevator spaces, two core optimizations are embedded into the network: the Programmable Gradient Information (PGI) training strategy, and a lightweight Generalized Efficient Layer Aggregation Network (GELAN) backbone enhanced with depthwise separable convolution (DSConv). A dedicated dataset consisting of roughly 2,000 images is established for model training and validation. This dataset covers a wide range of elevator scenes, diverse target postures, and common occlusion cases. Experimental results show that the proposed model achieves [email protected] values of 0.952, 0.915 and 0.887 across three challenging elevator scenarios. The average per-frame inference latency is less than 35 ms, which fully meets the real-time constraints for edge device deployment. To construct a complete closed-loop detection and alarm system, we further design an alarm subsystem empowered by multi-source information fusion. It adopts dual-threshold decision logic and supports multimodal alerts including audio, visual cues, and log recording. This framework first judges the presence of EMs via a basic confidence threshold, then verifies the temporal continuity of detection results using a dedicated alarm threshold, and finally triggers multimodal alarms accordingly. This system effectively improves the efficiency of alarm notifications and provides a practical technical solution for fire safety management of elevators in residential and commercial buildings.},
keywords = {electric motorcycles detection, YOLOv9, real-time system, intelligent alarm, multi-source information fusion},
issn = {2998-3371},
publisher = {Institute of Central Computation and Knowledge}
}
Article Metrics
Publisher's Note
ICCK stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and Permissions
Copyright © 2026 by the Author(s). Published by Institute of Central Computation and Knowledge. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.
Chinese Journal of Information Fusion
ISSN: 2998-3371 (Online)
| ISSN: 2998-3363 (Print)
Preserved at
Portico
Portico
Scan to read this article