Aiming at the problems of small object scale, complex background, dense target distribution and redundant detection branches in Unmanned Aerial Vehicle (UAV) aerial images, this paper proposes a lightweight UAV small object detection method based on YOLOv8n detection layer reconstruction. Instead of stacking complex enhancement modules, the proposed method adjusts the detection structure according to the matching relationship between target scale distribution and detection layers. A high-resolution P2 detection layer is introduced to strengthen shallow detail representation for small objects, while the P5 large-object detection layer is removed to reduce redundant parameters and computation. Thus, a lightweight P2-P3-P4 detection structure named YOLOv8-P2-woP5 is constructed. Experiments on the VisDrone2019 dataset show that the proposed model reduces parameters from 3.01M to 1.38M while maintaining an mAP50 of 0.332. Additional GhostConv experiments indicate that excessive compression may weaken small-object feature representation and reduce accuracy. The results demonstrate that the proposed model achieves a better balance between detection performance and model complexity.