MODERN APPROACHES TO MILITARY OBJECT RECOGNITION USING THE YOLOV8 MODEL
DOI:
https://doi.org/10.34132/mspc2025.01.14.16Keywords:
YOLOv8, neural networks, Python, dataset, markup, military objects.Abstract
The paper discusses modern approaches to military object recognition using the YOLOv8 neural network model. The main attention is paid to nano and small modifications, which provide an optimal balance between speed and accuracy. Training is carried out over 50 epochs based on a prepared dataset containing 12 classes of various military and civilian objects. The images were labeled and the models’ performance was analyzed under conditions close to real combat scenarios. The results obtained indicate the high suitability of lightweight versions of YOLOv8 for real-time defense and reconnaissance tasks. Special attention is paid to the possibility of integrating trained models into unmanned aerial vehicles (UAVs) for automatic target detection.
References
Alawi, M., & Mohammed, A. (2024). The Role of YOLOv8 in Enhancing Strategic Military Equipment Detection. ResearchGate.
Singh, S., & Ratna, G. N. (2024). Military Based Object Detection in Satellite Imagery by Optimising YOLOv8. IEEE Space, Aerospace and Defence Conference.
Alawi, M., & Mohammed, A. (2024). Empowering Military Vehicle Detection and Classification with YOLOv8 Model. ResearchGate.
Zhao, B., Zhou, Y., Song, R. et al. (2024) Modular YOLOv8 optimization for real-time UAV maritime rescue object detection. Sci Rep 14, 24492.
