Combination of systems for recognition and tracking of anomalous objects in telecommunication systems

Objective. The aim of the study is to integrate the YOLO and CSRT algorithms for automated recognition and tracking of objects in a video stream. The focus is on the problems associated with the growing volume of video data and the need for efficient identification of anomalous objects in security control systems. Method. The study is based on computer simulation methods and algorithms. Result. The key features of each algorithm, their advantages and disadvantages are defined. The results of experimental tests based on the analysis of video streams with various scenarios of object movement are presented, which show a reduction in frame processing time with high recognition accuracy. The possibility of further improvement of the automated recognition and tracking system is proven, in particular, adding the function of predicting the movement of objects, which will increase its efficiency and expand the scope of application in video surveillance and security tasks. Conclusion. Integration of algorithms allows achieving significant improvements in real time, which is especially important in the context of video surveillance and security. Adding the function of predicting the movement of objects will not only increase the functionality of the system, but also make it more adaptive to dynamic conditions, which is extremely important for preventing potential threats.

1. URL:www.researchgate.net/publication/376777967_Pepper_Target_Recognition_and_Detection_Based_on_Improved_YOLO_v4.Tan, Zhiyuan & Chen, Bin & Sun, Liying & Xu, Huimin & Zhang, Kun & Chen, Feng. Pepper Target Recognition and Detection Based on Improved YOLO v4.Information Technology and Control, 2023;12.

2. Farhadov, Xurshedjon & Lee, Suk-Hwan & Kwon, Ki-Ryong. Object Tracking using CSRT Tracker and RCNN.7th International Conference on Bioimaging. 2020:209-212.

3. Islam, Shahab & Ferraioli, Giampaolo & Pascazio, Vito & Vitale, Sergio & Amin, Muhammad. Performance Analysis of YOLOv3, YOLOv4 and MobileNet SSD for Real Time Object Detection // ResearchGate, 2024, № 6. – URL: www.researchgate.net/publication/381851712_Performance_Analysis_of_YOLOv3_YOLOv4_and_MobileNet_SSD_for_Real_Time_Object_Detection. 37-49.

4. Fang, Cheng & Huang, Jun duan & Cuan, Kaixuan & Zhang, Xiaolin & Zhang, Tiemin. Comparative study on poultry target tracking algorithms based on a deep regression network. // Biosystems Engineering, 2020. – URL: www.sci-hub.ru/10.1016/j.biosystemseng.2019.12.002

5. Shibaikin, S. D. Problems of constructing an optical flow of dynamic images / S. D. Shibaikin // Fundamental and applied problems of safety, survivability, reliability, stability and efficiency of systems: Proceedings of the III international scientific and practical conference dedicated to the 110th anniversary of the birth of academician N.A. Pilyugin, Yelets, June 03–05, 2019. Volume Part I. – Yelets: Yelets State University named after I.A. Bunin, 2019:81-88 (In Russ)

6. Guan, X. & Song, B. & Li, Z.. Comparison of the CSRT and the CST parameterization methods. Kongqi Donglixue Xuebao. Acta Aerodynamica Sinica. 2014;32:228-234.

7. Geng, Lei & Yang, Wenzhu & Jiao, Yanyan & Zeng, Shuang & Chen, Xinting. A multilayer human motion prediction perceptron by aggregating repetitive motion. // Machine Vision and Applications. 2023. 34.

8. Shibaikin S.D., D. A. Kryukova, D. A. Balykov, M. I. Shepelev, O. E. Shumarov Study of the application of tracking algorithms for determining and predicting deviant behavior. Scientific and Technical Bulletin of the Volga Region. 2023;12:635 - 638. (In Russ)

9. Mohit Phadtare, Varad Choudhari, Rushal Pedram and Sohan Vartak. Comparison between YOLO and SSD Mobile Net for Object Detection in a Surveillance Drone//IJSREM, 2021. – URL:www.researchgate.net/publication/355336797_Comparison_between_YOLO_and_SSD_MobileNet_for_Object_Detection_in_a_Surveillance_Drone.

10. www.researchgate.net/publication/344247798_Single_Object_Trackers_in_OpenCV_A_Benchmark. Adnan Brdjanin, Amila Akagic, Džemil Džigal, Nadja Dardagan. Single Object Trackers in OpenCV: A Benchmark // ResearchGate, 2020.

11. www.researchgate.net/publication/355222855_Multiple_Object_Trackers_in_OpenCV_A_Benchmark. Nadja Dardagan, Adnan Brdjanin, Džemil Džigal, Amila Akagic. Multiple Object Trackers in OpenCV: A Benchmark // ResearchGate, 2021.