Research on Underwater Target Detection Technology Based on SMV-YOLOv11n

High-precision underwater object detection is a crucial component for improving the efficiency of marine resource development. Traditional detection methods face challenges due to underwater environmental factors such as low brightness and high blurriness. To address these issues, this paper proposes an underwater object detection algorithm named SMV-YOLOv11. Firstly, the Swin Transformer is adopted to replace the backbone network, and a novel Multi-scale Spatial and Channel Attention module (MSCA) is designed and integrated into the backbone to enhance its feature extraction capability. Subsequently, the VoVGSCSP module is introduced to replace the C3K2 module, effectively leveraging contextual information to improve global detection performance. Furthermore, the WIoUv3 loss function replaces the CIoU loss function to enhance the model's robustness against spatial transformations and scale variations. Comparative experiments and ablation studies conducted on publicly available datasets validate the superiority and generalization ability of the improved model. SMV-YOLOv11 achieves an 87.6 % mean Average Precision (mAP), representing a 1.4 % improvement over the original YOLOv11. It also demonstrates significant gains of 2.7 %, 2.8 %, and 1.9 % in Precision, Recall, and F1 Score, respectively. The experimental results demonstrate that SMV-YOLOv11 possesses significant advantages in underwater object detection, effectively mitigating interference from background noise, low-light conditions, and blur. This offers a new solution for the field.