不同上下文比例对损毁建筑遥感场景图片样本集构建的影响

邰佳怡, 慎利, 乔文凡, 周吾珍

Impacts of different proportions of contextual information on the construction of sample sets of remote sensing scene images for damaged buildings

TAI Jiayi, SHEN Li, QIAO Wenfan, ZHOU Wuzhen

表1 分类实验结果表

Tab.1 Results of classification experiments(%)

模型 模型平均 上下文信息占比 70 75 80 85 90 95 AlexNet 90.27 90.40 90.40 90.52 90.63 90.40 89.28 VGGNet 92.18 92.42 92.20 92.59 92.42 92.14 91.30 GoogLeNet 93.13 93.99 94.22 93.99 93.21 92.20 91.19 ResNet 89.62 91.75 91.98 91.81 88.78 86.92 86.48 SqueezeNet 90.66 91.98 91.98 92.20 90.85 90.01 86.92 上下文信息占比平均 — 92.11 92.16 92.22 91.18 90.33 89.03