In response to the demand for target detection accuracy and real-time performance in complex environments for drone-based maritime search and rescue missions, research has been conducted to improve the YOLOv11 algorithm. By integrating Pinwheel-shaped Convolution to optimize the network backbone, designing a Feature Enhancement Module (FEM) and a Bidirectional Feature Pyramid Network (BiFPN) with adaptive weights, and introducing a dynamic attention mechanism, the approach achieves feature enhancement and noise suppression for small and occluded targets at sea. Experiments were conducted using the SeaDronesSee dataset, and the results indicate that the improved model achieved a detection accuracy (mAP@0.5) of 78.47% and an inference speed (FPS, Frames Per Second) of 511.79 Hz, outperforming traditional YOLO series algorithms. Validation through practical applications shows that this algorithm can meet the high accuracy and real-time requirements of maritime search and rescue missions, providing effective technical support for intelligent emergency rescue.