Prompt identification of low-altitude drones is vital for safeguarding personal privacy and deterring unauthorized intrusion. Therefore, an enhanced YOLO11n-based visual detection model is introduced. Firstly, a BiSAFPN is developed and presented, capable of concurrently acquiring and integrating global and local feature information across various scales to boost detection precision and strengthen model robustness; second, the RCS-OSA is introduced, and the channel mixing technology of the module enables the feature information of different channels to communicate better and capture the details and contextual information of the target better; Finally, the original detection head is replaced by the DSDH, and through the task alignment mechanism, the classification and localization tasks share features and enhance each other to boost detection accuracy and adaptability to various scenes and tasks. The findings demonstrate that, compared with the original model, the enhanced model achieves improvements of 6.2%, 6.5%, 6.6%, and 3.6% in P, R, mAP50, and mAP50:95, respectively, on the YOLO Drone Detection Dataset sourced from Kaggle. This demonstrates the effectiveness of the improved model.