Accurate detection of underwater docking stations and their 3D pose estimation for autonomous underwater vehicles (AUVs) is a very challenging task due to various factors such as turbid water, light scattering, low visibility in deep oceanic regions, environmental pollution, etc. To overcome these challenges, the paper focuses on developing a detection algorithm capable of excelling in adverse underwater conditions. The paper leverages state-of-the-art deep-learning detection algorithms, including YOLOv8, YOLOv11 and Detection Transformer. From the experiments, it has been found that RT-DETR works better than YOLOv8 and YOLOv11 in recognizing underwater docking stations, as seen by its flawless recall and better classification loss. While both YOLOv8 and RT-DETR models have comparable mAP50 scores, YOLOv11 performs somewhat better, at an IoU of 0.5. Moreover, YOLOv11 performs exceptionally well in mAP50-95, highlighting its general proficiency over various IoU thresholds.