Штучний інтелект

Анотація: The article explores the theoretical and practical aspects of designing a video analytics complex, the purpose of which is to continuously monitor and track visual targets in dynamic scenes. A thorough critical review of software solutions and open algorithmic paradigms available on the market has been carried out, which has allowed us to argue the need for the development of optimized tools. The foundation of the developed system is the YOLOv8m convolutional archi-tecture, adapted for localization of entities in complex environmental conditions, including changes in illumination, scaling of the target and its short-term overlap with other objects (occlusion). The mathematical apparatus underlying the algorithm for frame-by-frame estimation of the approximated speed of movement of detected elements is detailed, and the mechanisms for saving their identifiers are also disclosed. An intuitive graphical user interface has been de-signed, which provides the possibility of interactive processing of both pre-saved media files and streaming video data from webcams in real time. The results obtained and the software product have potential for integration into security systems, traffic monitoring systems, and can also serve as a basic visual navigation module for unmanned aerial vehi-cles.

Ключові слова: computer vision, object tracking, convolutional neural networks, YOLO architecture, video analytics, detection algo-rithms, unmanned aerial vehicles, mathematical modeling, situation centers.

Посилання:

1. Pysarenko, Y. V., Melkumyan, K. O., Varava, I. V., Koval, O. S., & Chumakova, N. F. (2022). On the organization of regional situational centers of the intellectual system "CONTROL_TEA" using UAVs. Artificial Intelligence, (1), PP. 275-281. https://doi.org/10.15407/jai2022.01.275
2. Pisarenko, J., & Melkumyan, E. (2019). The structure of the information storage “CONTROL_TEA” for UAV applications. IEEE 5th International Conference on Actual Problems of Unmanned Aerial Vehicles Developments (APUAVD), 274–277. https://doi.org/10.1109/APUAVD47061.2019.8943938.
3. V. G. Pisarenko, N. V. Nogin, A. S. Kryachok, J. V. Pisarenko, I. A. Varava, A. S. Koval. (2022) About complex intelligent technologies for techno-ecological events control in the water area. Prombles in programming, 3-4, PP. 437-445. https://doi.org/10.15407/pp2022.03-04.437
4. V. G. Pisarenko, J. V. Pisarenko, O. E. Gul-chak, T. I. Chobotok, A. G Boyko. (2021) Practical experience in the technical systems creating with the artificial intelligence elements. Artificial Intelligence, (1), PP. 95-101. https://doi.org/10.15407/jai2021.01.095
5. Pysarenko Yu. V., Karmazin K. V. Vy-korystannia ShI dlia obrobky ta analizu velykykh obsiahiv danykh u realnomu chasi z intehratsiieiu multyahentnykh system // Artificial Intelligence. – 2025. – № 4. – PP.134-145 https://doi.org/10.15407/jai2025.04.134
6. Bewley, A., Ge, Z., Ott, L., Ramos, F., & Up-croft, B. (2016). Simple online and realtime tracking. In 2016 IEEE International Conference on Image Processing (ICIP). – PP. 3464-3468. https://doi.org/10.1109/ICIP.2016.7533003