The integration of Unmanned Aerial Vehicles (UAVs) with advanced computer vision techniques has revolutionized power line inspection, offering enhanced safety and efficiency. This paper presents a novel approach utilizing the cutting-edge YOLOv8 image segmentation algorithm with the Realsense D455 depth camera for precise 3D localization of power line components. We incorporate the YOLOv8 image segmentation technique into the extraction of depth maps for power lines, and have developed a novel method that utilizes histogram analysis to eliminate noise from segmented depth maps. Furthermore, we propose a simplified approach for fitting power lines. Our methodology showcases robust and real-time enhancements in detection and localization accuracy. We validate our approach through two experiments: the first quantify the precision of pose estimations against ground truth data from a motion capture system, and the second showcases the practical application via a robotic arm’s successful contact with power lines, guided by pose estimations. The results indicate the system’s potential for automated power line inspection and maintenance. Code is available at https://github.com/4399chen/choosepowerline .