To address the issues of low detection accuracy and false or missed detection of small-sized drones in complex flying environments such as schools and parks, a small target drone detection method based on the improved YOLOv8n is proposed. Firstly, drone images in different flight backgrounds are collected to build an experimental dataset. Secondly, the multi-scale feature fusion network is redesigned, introducing TPE and SSFF modules to improve the multi-scale feature fusion method of the neck network, and a small target detection layer is added to enhance the network ability to resist background interference and the detection accuracy for small targets. Finally, Inner-CIoU is used as the loss function of the improved model to enhance the model detection accuracy and convergence speed. Experimental results on the self-built drone Anti-Drone dataset show that compared with YOLOv5s, YOLOv7-tiny, YOLOv7, and YOLOv8s algorithms, the proposed method increases the value of mAP50 by 0.8, 15.5, 9.8, and 5.2 percentage points respectively, which demonstrates the effectiveness of the improved method in detecting small-scale drones in complex backgrounds.