Comparative Analysis of YOLO V8, V9, and V10 for Multi-Target Detection in Military Applications

The ability of military vision systems to accurately identify multiple targets is crucial, especially for autonomous defense and surveillance applications. This capability is particularly important for autonomous defense applications. This research conducts a comparative analysis of the YOLO V8, V9, and V10 models, utilizing 18K-image data from nine object classes pertinent to military functions. YOLO V9 achieved the best overall average accuracy, scoring 88.11%. Regarding accuracy (96.6% for V10 and 98.60% for V8 and V9) and mAP@ as return metrics, V8 and V10 provide results of 86.66% and 86.55%, respectively. These models are reliable because of measurements like the mAP@ and F1 score. The camouflage soldier class gave us a challenge, and the HUMVEE class had the best detection accuracy. We utilized real-time testing and live video feeds to investigate the potential military applications of these models, such as border surveillance and integration with autonomous vehicles. After verifying these three models, our findings were satisfactory.