重构YOLOv11的无人机小目标检测算法
孟昱煜,孔垂乐,火久元,武泽宇

UAV small target detection algorithm based on reconstruction of YOLOv11
Yuyu MENG,Chuile KONG,Jiuyuan HUO,Zeyu WU
表 2 VisDrone2021数据集上不同算法的性能对比实验结果
Tab.2 Performance comparison of different algorithms on VisDrone2021 dataset
模型尺寸算法P/%R/%mAP50/%mAP50-90/%Np/106FPS/(帧·s−1)FLOP/(109·s−1)
NDrone-YOLO46.535.636.121.33.079.412.4
YOLOv5n42.832.032.318.22.296.65.8
YOLOv6n39.831.230.317.54.297.811.8
YOLOv8n44.231.932.419.03.0101.58.1
YOLOv8-Ghostp244.032.332.618.81.666.87.3
YOLOv9t44.033.133.319.32.071.87.6
YOLOv10n43.032.432.518.72.770.28.2
YOLOv11n42.733.032.918.82.682.36.3
YOLOv12n41.731.430.917.82.546.76.0
DLSRF-Net49.938.140.124.22.8109.811.9
MYOLOv8s49.536.937.822.611.298.628.5
YOLOv9s50.038.739.423.47.236.726.7
YOLOv10s48.037.537.922.58.168.924.5
YOLOv11s48.237.737.922.79.479.621.3
YOLOv12s48.536.836.922.09.147.819.3
RT-DETR54.442.745.327.519.844.836.7
DLSRF-Net60.145.647.328.67.1103.125.5