Elevator entrapment incidents are frequent precursors to serious elevator accidents, posing significant risks to passenger safety. However, most existing elevator monitoring systems rely on post-event analysis and lack the ability to proactively identify abnormal passenger behaviors, especially under the confined, occluded, and visually complex conditions of elevator cabins. This poses a challenge for developing accurate and lightweight vision-based early warning methods suitable for real-time deployment. To address this issue, this paper proposes an elevator entrapment early warning action recognition method based on an improved YOLOv8 classification model, termed YOLOv8-DA. A dedicated elevator behavior dataset containing 5,501 images is constructed, covering six typical passenger behaviors related to entrapment risk. In addition, a Dual Attention Fusion Module (DAFM), integrating Efficient Channel Attention Network (ECA-Net) channel attention and Coordinate Attention mechanisms, is embedded into the YOLOv8-cls backbone to enhance both global semantic representation and fine-grained spatial discrimination. Experimental results demonstrate that YOLOv8-DA achieves an accuracy of 97.18% with only 2.7 M parameters and an inference speed of 116 FPS, outperforming representative lightweight and classical classification models. The proposed method provides an effective and practical solution for proactive elevator entrapment early warning and edge-side deployment.
References
[1] Prahlow, J. A., Ashraf, Z., Plaza, N., Rogers, C., Ferreira, P., Fowler, D. R., Lantz, P. E. (2020) Elevator-related deaths. Journal of Forensic Sciences, 65(3), 823-832.
[2] Zhenbo, C., Mu, Y., Haoxin, Z., Xuesong, X., Wenchao, L., Kun, S., Gang, X. (2025) Review of fault modes, diagnosis and prediction methods for elevator systems. Journal of Vibration Engineering & Technologies, 13(7), 473.
[3] Ali, M. L., Zhang, Z. (2024) The YOLO framework: a comprehensive review of evolution, applications, and benchmarks in object detection. Computers, 13(12), 336.
[4] Zhang, C., Li, Z., Li, J., Zou, L., Dong, E. (2025) Optimization of visual detection algorithms for elevator landing door safety-keeper bolts. Machines, 13(9), 790.
[5] Luo, J., Yang, X., Dai, Q., Qiu, W., Nie, S., Wu, J., Zeng, M. (2025) Multimodal fusion-based self-calibration method for elevator weighing towards intelligent premature warning. Sensors, 25(17), 5550.
[6] Wang, Z., Chen, J., Yu, P., Feng, B., Feng, D. (2024) SC-YOLOv8 network with soft-pooling and attention for elevator passenger detection. Applied Sciences, 14(8), 3321.
[7] Liu, H., Zhang, Y., Chen, Y. (2024) A symmetric efficient spatial and channel attention (ESCA) module based on convolutional neural networks. Symmetry, 16(8), 952.
[8] Guo, Y., Liu, Y., Georgiou, T., Lew, M. S. (2018) A review of semantic segmentation using deep neural networks. International Journal of Multimedia Information Retrieval, 7(2), 87-93.
[9] Khan, A., Sohail, A., Zahoora, U., Qureshi, A. S. (2020) A survey of the recent architectures of deep convolutional neural networks. Artificial Intelligence Review, 53(8), 5455-5516.
[10] Zhang, S., Liu, Z., Chen, Y., Jin, Y., Bai, G. (2023) Selective kernel convolution deep residual network based on channel-spatial attention mechanism and feature fusion for mechanical fault diagnosis. ISA Transactions, 133, 369-383.
[11] Altuwaijri, G. A., Muhammad, G., Altaheri, H., Alsulaiman, M. (2022) A multi-branch convolutional neural network with squeeze-and-excitation attention blocks for EEG-based motor imagery signals classification. Diagnostics, 12(4), 995.
[12] Yang, J., Li, C., Zhang, P., Dai, X., Xiao, B., Yuan, L., Gao, J. (2021) Focal attention for long-range interactions in vision transformers. Advances in Neural Information Processing Systems, 34, 30008-30022.
[13] Hassan, H. Z., Saeed, N. M. (2024) Advancements and applications of lightweight structures: a comprehensive review. Discover Civil Engineering, 1(1), 47.
[14] Nguyen, H. (2020) Fast object detection framework based on mobilenetv2 architecture and enhanced feature pyramid. J. Theor. Appl. Inf. Technol, 98(05), 812-824.
Share and Cite
Li, H., Hu, H., Yang, S. (2025) Research on Elevator Entrapment Early Warning Action Recognition Method Based on YOLOv8-DA. Scientific Research Bulletin, 2(6), 18-28.