In order to solve the problem of high missed detection rate and false detection rate in human detection for drone aerial images, this paper proposes a human detection algorithm for aerial images based on improved YOLOv8. First, the C2f-pcp module was designed to lightweight the model by replacing the Bottleneck network in C2f with the FasterNet-pcp network. Secondly, the original CIoU loss function of YOLOv8 is replaced with WIoUv3 to improve the detection accuracy of the model. Finally, the network structure is improved by adding small-scale target detection head, deleting large-scale target detection head and introducing BIFPN design to improve the model's small target detection ability and feature fusion ability. Experiments were conducted on the SARD data set. The results show that the average detection accuracy of this algorithm is 3.4% higher than YOLOv8n, reaching 81.3%, and has better detection results in emergency rescue scenarios.