基于多尺度特征增强的航拍小目标检测算法
肖剑,何昕泽,程鸿亮,杨小苑,胡欣

Aerial small target detection algorithm based on multi-scale feature enhancement
Jian XIAO,Xinze HE,Hongliang CHENG,Xiaoyuan YANG,Xin HU
表 5 不同算法在VisDrone数据集上的平均精度和参数量对比
Tab.5 Comparative results of different algorithms in average precision and parameters on VisDrone 2019 dataset
模型AP/%mAP0.5/%F/(帧$ \cdot {{\mathrm{s}}^{ - 1}}$)Para/106
行人自行车汽车面包车卡车三轮车遮阳棚三轮车巴士摩托车
Faster R-CNN[20]20.914.87.351.029.719.514.08.830.521.221.814.4
YOLOv5s39.031.311.273.535.429.520.511.143.137.033.2118.07.03
TPH-YOLOv5[1]53.342.121.183.745.242.533.016.361.151.044.934.060.42
YOLOv7-tiny[21]37.934.69.476.136.329.820.110.643.241.834.089.06.03
YOLOv8s42.032.812.579.444.735.526.917.154.043.338.817.311.13
YOLOv8l51.139.821.982.949.345.438.120.467.052.346.859.043.69
YOLOv9-C[22]34.018.415.477.545.254.124.824.164.938.339.750.90
YOLOv11s41.631.811.279.545.435.526.115.555.143.338.5121.89.46
本研究模型53.9 48.327.581.353.649.340.024.571.952.850.357.38.08