标题： 基于卫星图像和深度学习的非洲基础设施质量评估 显示英文标题

标题： Infrastructure Quality Assessment in Africa using Satellite Imagery and Deep Learning

摘要： 联合国可持续发展目标提到了基础设施质量在十七个目标中的三个的重要性。 然而，在发展地区监测基础设施质量仍然成本高昂，并阻碍了衡量这些目标进展的努力。 为此，我们研究了使用广泛可用的遥感数据来预测非洲的基础设施质量。 我们训练了一个卷积神经网络，从Afrobarometer第六轮调查中预测真实标签，使用Landsat 8和Sentinel 1卫星图像。 我们最好的模型在电力方面AUROC得分为0.881，在下水道方面为0.862，在管道供水方面为0.739，在道路方面为0.786，使用Landsat 8。 这些性能明显优于利用OpenStreetMap或夜间灯光强度的模型。 我们还证明，在对少量图像进行微调后，我们的训练模型可以在未见过的国家中准确进行预测。 此外，该模型可以部署在样本有限的地区，以比最近邻空间插值更高的性能预测基础设施结果。 显示英文摘要

摘要： The UN Sustainable Development Goals allude to the importance of infrastructure quality in three of its seventeen goals. However, monitoring infrastructure quality in developing regions remains prohibitively expensive and impedes efforts to measure progress toward these goals. To this end, we investigate the use of widely available remote sensing data for the prediction of infrastructure quality in Africa. We train a convolutional neural network to predict ground truth labels from the Afrobarometer Round 6 survey using Landsat 8 and Sentinel 1 satellite imagery. Our best models predict infrastructure quality with AUROC scores of 0.881 on Electricity, 0.862 on Sewerage, 0.739 on Piped Water, and 0.786 on Roads using Landsat 8. These performances are significantly better than models that leverage OpenStreetMap or nighttime light intensity on the same tasks. We also demonstrate that our trained model can accurately make predictions in an unseen country after fine-tuning on a small sample of images. Furthermore, the model can be deployed in regions with limited samples to predict infrastructure outcomes with higher performance than nearest neighbor spatial interpolation.