LA-YOLO: an effective detection model for multi-UAV under low altitude background

Accurately detecting unmanned aerial vehicle (UAV) is highly demanded in various fields, for solving the problem of detecting UAV under low altitude background, we propose the LA-YOLO network, which achieves satisfactory performance by combining the SimAM attention mechanism, introducing an effective fusion block and mixing the normalized Wasserstein distance. Through taking images of UAV under low altitude background and annotating them, we constructed a dataset UAV-LA to evaluate the performance of the proposed method. Experiments validate the good performance of the proposed method on both the homemade dataset and publicly available datasets, which guarantees the high stability for surveillance of the multi-UAV using visible light images.