论文标题

脉冲显示:一种使用人道主义背景下卫星图像分析的人类反馈回路的工具

PulseSatellite: A tool using human-AI feedback loops for satellite image analysis in humanitarian contexts

论文作者

Logar, Tomaz, Bullock, Joseph, Nemni, Edoardo, Bromley, Lars, Quinn, John A., Luengo-Oroz, Miguel

论文摘要

卫星图像分析可以帮助人道主义对自然灾害和冲突的反应。在人道主义背景下,必须及时地准确地完成非常具体的卫星图像分析任务,以提供运营支持。我们提出了脉冲星是一种协作的卫星图像分析工具,该工具利用了可以随时训练并适应特定人道主义背景和地理位置的神经网络模型。我们分别在映射避难所和洪水中提出了两个案例研究,这些案例研究说明了脉冲黄石的能力。

Humanitarian response to natural disasters and conflicts can be assisted by satellite image analysis. In a humanitarian context, very specific satellite image analysis tasks must be done accurately and in a timely manner to provide operational support. We present PulseSatellite, a collaborative satellite image analysis tool which leverages neural network models that can be retrained on-the fly and adapted to specific humanitarian contexts and geographies. We present two case studies, in mapping shelters and floods respectively, that illustrate the capabilities of PulseSatellite.

扫码加入交流群

加入微信交流群

微信交流群二维码

扫码加入学术交流群,获取更多资源