Witnessing atrocities: quantifying villages destruction in Darfur with crowdsourcing and transfer learning

Published in NeurIPS AI for Social Good Workshop, 2018

Abstract

We report the first, to the best of our knowledge, hand-in-hand collaboration between human rights activists and machine learners, leveraging crowdsourcing and transfer learning to automatically analyze satellite imagery, at country-wide scale, for conflict reporting. This will allow for large-scale quantitative evidence gathering, beyond closed borders, usable in conjunction with survivors testimonials to demand justice and accountability.

  1. @article{CornebiseWFM18,
      author = {Cornebise, Julien and Worrall, Daniel E. and Farfour, Micah and Marin, Milena},
      title = {Witnessing atrocities: quantifying villages destruction in Darfur with crowdsourcing and transfer learning},
      journal = {NeurIPS AI for Social Good Workshop},
      year = {2018}
    }