Ensuring safe operations on busy airport aprons remains a significant challenge, especially in preventing aircraft wingtip collisions. Firstly, a simplified coordinate mapping technique is introduced to convert pixel-based detections into accurate spatial coordinates, thereby enhancing estimates of aircraft position and velocity. This study introduces a real-time detection algorithm based on YOLOv8-Pose, which is enhanced through lightweight modules to significantly reduce model size and computational overhead while maintaining accuracy. Experimental results on a real Airport dataset, representing various apron configurations, demonstrate frame rates of up to 461.7 FPS and a 90.5\% reduction in model size compared to the baseline. Furthermore, static and dynamic alert zones are designed around the wingtips to capture both stationary and taxiing aircraft conflicts, achieving a detection success rate exceeding 97\%. Visualizations further demonstrate that the solution is versatile and high-speed, effectively mitigating wingtip collisions and enhancing apron safety.Keywords:collaborative perception; YOLOv8-Pose; coordinate-conversion; Pose-LSCD; computing resources