基于多尺度特征聚合的航拍图像检测算法
李珺,丁彬彬,史维娟,杨琳

Aerial image detection algorithm based on multiscale feature aggregation
Jun LI,Binbin DING,Weijuan SHI,Lin YANG
表 6 不同算法的检测性能对比
Tab.6 Comparison of detection performance of different algorithms
模型P/%R/%mAP50/%mAP50-
95/%
Np/
106
FLOPs/
109
YOLOv8n42.932.932.819.03.08.1
YOLOv8s49.638.138.623.111.128.5
YOLOv8m53.341.642.726.025.878.7
YOLOv10n43.833.333.219.02.36.5
YOLOv10s49.939.039.423.37.221.4
YOLOv10m53.641.642.925.915.358.9
YOLOv11n42.732.732.310.72.66.3
YOLOv11s49.737.538.623.19.421.3
YOLOv11m54.842.043.826.820.067.7
YOLOv12n41.832.431.618.125.15.8
YOLOv12s49.536.537.622.59.119.3
YOLOv12m53.240.742.225.719.659.5
YOLOv13n42.931.531.418.12.46.1
YOLOv13s47.637.237.122.19.020.1
YOLOv13l53.641.942.525.826.984.4
Hyper-YOLOn44.835.135.020.63.69.5
Hyper-YOLOs51.238.939.824.013.533.8
Hyper-YOLOm53.741.542.526.030.791.8
RT-DETR-r1858.541.543.826.619.857.0
RT-DETR-r3461.545.547.229.031.188.8
AAPW-YOLOn[18]49.937.338.622.42.111.7
PC-YOLOn[19]46.835.436.121.51.9
YOLO-S3DTn[20]47.437.937.921.12.911.0
Eagle-YOLOs[21]53.643.342.925.016.6
PARE-YOLOs[22]60.645.446.328.4
RPS-YOLO[2]55.344.346.328.113.337.7
FDL-YOLOn49.239.339.924.11.110.2
FDL-YOLOs55.645.747.229.12.928.9