基于SU-RetinaNet的高分辨率遥感影像非正规垃圾堆检测

吴同, 彭玲, 胡媛

Informal garbage dumps detection in high resolution remote sensing images based on SU-RetinaNet

WU Tong, PENG Ling, HU Yuan

表1 不同参数配置下RetinaNet在验证集AP对比

Tab.1 Comparison of AP on the validation set under different parameter configurations using RetinaNet (%)

阶段 特征提取网络深度 α=0.25, γ=1 α=0.25, γ=2 α=0.25, γ=5 α=0.5, γ=1 α=0.5, γ=2 α=0.5, γ=5 α=0.75, γ=1 α=0.75, γ=2 α=0.75, γ=5 阶段1 ResNet50 68.26 74.47 76.88 75.71 75.03 79.38 72.62 75.98 75.63 ResNet101 72.52 72.09 74.25 70.84 72.49 78.78 70.93 77.09 75.46 ResNet152 75.28 77.48 68.85 73.33 73.84 74.80 71.65 76.22 74.98 阶段2 ResNet50 81.39 82.34 84.54 82.93 85.18 85.38 84.28 87.25 85.98 ResNet101 81.49 83.76 83.17 78.99 83.77 83.59 79.97 82.03 83.26 ResNet152 79.89 82.6 79.32 80.35 84.33 83.18 77.7 82.41 84.08