Research on Small Target Detection Algorithm for Aerial Images based on Improved YOLOv11

Abstract

At present, UAV aerial photography has a good application prospect in agricultural production and disaster response. The application of drones can greatly improve work efficiency and decision-making accuracy. However, due to the inherent characteristics of drone aerial images, such as high image density, small target size, complex background, etc. In order to solve these problems, this paper proposes a small target detection algorithm for UAV aerial photography based on the improved YOLOv11n. Firstly, the FADC module was introduced into the backbone network to optimize the feature extraction process. Then, a small target detection layer was introduced into the algorithm to improve the detection performance of small targets in aerial images. Secondly, the scale sequence feature fusion network ASF-YOLO was used to replace the PANet network to improve the speed and accuracy of target detection. Then, Wise IoU is used to replace CIoU to speed up the network convergence speed and improve the regression accuracy. The algorithm was evaluated on the VisDrone-2019 dataset. Compared with YOLOV11n, the algorithm is improved by 5.7% and 4.3% in mAP@50 and [email protected]:0.95, respectively. Experiments show that compared with YOLOV11n, the performance of the algorithm on small targets is greatly improved.