The approach phase is a critical stage in flight operations. Using electro-optical equipment to monitor aircraft can visually identify the risk of deviation from the glideslope, providing a reliable basis for controllers to intervene, which is of great significance for ensuring flight safety. In this paper, a monocular vision method based on improved YOLOv8n was proposed to determine aircraft deviation from the glideslope. Based on the YOLOv8n network, the MobileNetV4 network was adopted as the backbone, reducing the number of model parameters and computational complexity while maintaining strong feature extraction capability. A strip attention block was designed to enhance the perception of directional features of slender targets. A centroid correction module and a simplified SORT algorithm were integrated to obtain stable, accurate, and continuous aircraft flight trajectories. The trajectories were then intersected with a ray-plane to calculate the distance and height of the aircraft from the runway threshold, which were compared with the theoretical glideslope equation to determine whether the aircraft deviated from the glideslope. Experimental results show that in terms of object detection, the YOLOv8n-MBSA model achieves 97.7% average precision on the self-built aircraft approach dataset, while the number of model parameters is only 1.83 M. Both detection accuracy and computational load outperform the original YOLOv8n model. In terms of object tracking, the tracking speed reaches 44.35 FPS, and a tracking success rate of 90.3% is achieved while meeting real-time tracking requirements. The calculated flight trajectory differs little from the theoretical glideslope, providing a reliable basis for judging glideslope deviation. This study provides an efficient and feasible technical solution for determining whether an aircraft deviates from the glideslope during the approach phase using electro-optical equipment.