 This paper proposed a lightweight object detection algorithm for remote sensing images. It replaces the existing YOLOV5S architecture with a new DDHED module, SPPC SPG module, SA module, CARAF module, and GSCONV module. This combination of modules improved the detection accuracy of the algorithm by 1.4 percent and 1.2 percent on two remote sensing data sets and 1.4 percent and 3.1 percent on two conventional object detection data sets. This article was authored by Pengfei Liu, Qing Wang, Huan Zhang, and others.