AI-Driven Defence Surveillance and GPS Threat Alert System based on YOLO Detection for Land, Naval and Airspace Monitoring

This article introduced AIDS-GTAS, a multi-domain defense surveillance and GPS threat alert system powered by AI, integrated with YOLOv11 with SAHI-based small-object enhancement, GPS georeferenced localization, and autonomous Telegram notification across land, naval, and airspace surveillance domains. The experimental results on the 18,400-frame CDSD-2026 dataset show that the proposed YOLOv11+SAHI configuration achieved [email protected] of 94.7%, Precision of 96.2%, Recall of 93.8% and 47.3 FPS inference speed, outperforming all the baselines evaluated. The SAHI ablation study quantified 11.3 percentage point mAP improvement for extra small objects. The GPS alert pipeline achieved a mean end-to-end latency of 1.84 seconds, a 73% improvement over human-operator benchmarks reported in the literature for operational conditions.   The architecture’s modularity, the feasibility of edge AI deployment on NVIDIA Jetson platforms, and the free Telegram notification infrastructure make AIDS-GTAS a practically deployable solution for resource constrained defense facilities. Use: perimeter protection of military bases, border monitoring, coast monitoring, verification of airspace integrity, protection of critical infrastructure. Future work includes coupling of thermal infrared, federated learning, and autonomous counter-UAV response .

[1]  J. B. White and M. L. Stevens, "Historical evolution of military surveillance technology," IEEE Trans. Aerosp. Electron. Syst., vol. 58, no. 4, pp. 3112–3128, Aug. 2022. [2]  R. Kumar and P. Sharma, "Commercial UAV threat proliferation in modern conflict environments," J. Defence Technol., vol. 19, no. 2, pp. 144–162, Apr. 2023. [3]  A. L. Thompson, D. R. Harkins, and S. P. Nguyen, "Video data volume analysis in brigade-level surveillance networks," Proc. IEEE Int. Symp. Technol. Homeland Secur., pp. 1–7, Nov. 2021. [4]  K. He, X. Zhang, S. Ren, and J. Sun, "Deep residual learning for image recognition," Proc. IEEE CVPR, pp. 770–778, 2016. [5]  J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, "You only look once: Unified, real-time object detection," Proc. IEEE CVPR, pp. 779–788, 2016. [6]  S. Ren, K. He, R. Girshick, and J. Sun, "Faster R-CNN: Towards real-time object detection with region proposal networks," IEEE Trans. PAMI, vol. 39, no. 6, pp. 1137–1149, Jun. 2017. [7]  M. D. Rao and V. C. Prasad, "Cognitive fatigue in prolonged video surveillance tasks," Hum. Factors, vol. 64, no. 3, pp. 487–503, May 2022. [8]  F. Zhu et al., "Learning spatial regularization for multi-label image classification," IEEE Trans. Image Process., vol. 30, pp. 2582–2594, 2021. [9]  Y. Chen, C. Li, X. Guo, and J. Jia, "Multi-domain adaptation for object detection in surveillance," Proc. IEEE ICIP, pp. 2741–2745, 2022. [10]   S. E. Mostafa, A. Elngar, and N. Ghali, "Impact of meteorological conditions on deep learning surveillance," IEEE Access, vol. 10, pp. 78432–78451, 2022.