Ship Detection Algorithm Based on Improved YOLO11

Aiming at the problems of high rates of missed detection and false detection in existing ship detection algorithms, YOLO-Mamba-Ship algorithm based on the YOLO11 is proposed for ship detection. Firstly, in the C3K2 module of the backbone network, the EMA (Efficient Multi-Scale Attention) mechanism is introduced to enhance the algorithm's ability to perceive the target location information with multi-scale differences. Secondly, in neck network, the VSS module is designed, integrating the SSM (State Space Model) to capture long-distance dependencies, which alleviates the notable problems that CNNs (Convolutional Neural Networks) are difficult to capture long-distance dependencies and Transformer models have quadratic computational complexity. The experimental results demonstrate that the proposed algorithm YOLO-Mamba-Ship improves the precision by 2.8%, the recall by 6.2%, and the F1 score by 4.7%, the mAP@0.5 by 7.2%, the mAP@0.5:0.95 by 5.3% compared with the baseline algorithm YOLO11, which improves the detection accuracy of the ship significantly and enhances robustness.