Journal of Engineering, Project, and Production Management, 2026, 16(3), 2025-221
Improved YOLO-Based Abnormal Behavior Recognition System for Civil Aviation Security
1
Professional Instructor, Airport Management College, Guangzhou Civil
Aviation College, Guangzhou 510403, China
2 Instructor, Airport Management
College, Guangzhou Civil Aviation College, Guangzhou 510403, China,
E-mail: tianyouwuty@126.com (corresponding author).
Project Management
Received October 11, 2025; revised November 26, 2025; accepted December 11, 2025
Available online April 8, 2026
Abstract: As the global aviation industry continues to expand rapidly, the nature of security threats has grown more intricate. Under the influence of the international security situation, aviation screening systems have emerged as a key defense layer. In response, this paper puts forward a composite model that integrates multiple algorithms. First, the model combines a focal loss function, which is suitable for detecting dense targets, with a target detection algorithm. Following that, the particle swarm optimization algorithm parameters are fine-tuned using a Bayesian optimization approach. Finally, the model integrates the focal loss-based hybrid target detection algorithm with the optimized particle swarm optimization algorithm to construct a behavior recognition model. Experimental results show that this model's loss curve is smoother and has lower loss than the three traditional models. On the University of Central Florida dataset and the University of Rochester dataset, the model achieves mean average precision values of 97.675% and 98.246%, respectively, outperforming the other three models. In addition, the model reaches a maximum accuracy of 98.652% and a maximum FPS of 140 on two aviation screening datasets. These results demonstrate that the model offers both high accuracy and real-time performance, indicating its potential to support the development of aviation screening systems and effectively meet the demands of security screening.
Keywords: YOLO, abnormal behavior detection, aviation security, deep learning, real-time recognition.
Copyright © Journal of Engineering, Project, and Production Management (EPPM-Journal).
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License.
Requests for reprints and permissions at eppm.journal@gmail.com.
Citation: Chen, F. and Wu, T. (2026). Improved YOLO-Based Abnormal Behavior Recognition System for Civil Aviation Security. Journal of Engineering, Project, and Production Management, 16(3), 2025-221.
