Remote detection of village burnings in Darfur

the solution proposed

Project motivation

On April 15, 2023, fighting in Sudan’s capital, Khartoum, between the Sudan Armed Forces (SAF) and the Rapid Support Forces (RSF), rapidly spread throughout the country. SAF leader Gen. Abdelfattah al-Burhan and RSF leader Gen. Mohamed Hamdan Dagalo, known as “Hemedti” had jointly carried out a coup against the country’s transitional government in October 2021. Both forces have a history of violating international humanitarian and human rights law, notably in Darfur and during crackdowns against protesters. In 2022, the trial of Ali Kosheib, a former Janjaweed militia leader, charged with 31 counts of war crimes and crimes against humanity, began at the International Criminal Court. Arrest warrants against ex-president Omar al-Bashir and two of his associates remain outstanding. More information by Human Rights Watch on the conflict can be found here and here.

Data

We make use of two main datasets:

  • Satellite data on fire by NASA FIRMS, which covers the whole globe roughly every three hours
  • Settlement data by Grid3 also obtained from satellite images (using semantic segmentation)

Telegram Bot

As part of the project we developed a Telegram Bot that allows to keep track of burning settlements. We developed an alert score to flag the most relevant ones. The Telegram Bot can be found under darfur_bot or here. In order to monitor the Darfur region consistently we created a Telegram channel that provides daily updates on the flagged settlements. At this point the Telegram channel is private and access can be granted individually upon request.

Marcello Negri
Marcello Negri
PhD candidate

I am a PhD student in machine learning currently trying to make models more flexible and interpretable.