World Food Programme (2020) Report (multilateral / development partner)

The Benefits of Machine Learning in Emergencies: From River DEEP to Mountain SKAI

World Food Programme

Ref: SRC-003-MOZ-002

https://drones.wfp.org/blog/benefits-of-machine-learning-in-emergencies

Accessed: 10/31/2025

Summary

WFP blog post describing the evolution of DEEP from drone-based to satellite-based analysis (SKAI). Reports DEEP achieves 85% accuracy in pinpointing damaged buildings and reduces assessment time from weeks to less than one hour. Confirms DEEP was first deployed in Mozambique 2019, subsequently in Colombia, Philippines, and Lebanon in 2020. Notes DEEP can run on any laptop with suitable video card and operates offline.

View Harvard reference

World Food Programme (2020) 'The Benefits of Machine Learning in Emergencies: From River DEEP to Mountain SKAI', WFP Drones. Rome: WFP. Available at: https://drones.wfp.org/blog/benefits-of-machine-learning-in-emergencies (Accessed: 31 October 2025).

Linked Cases (1)

MOZ-002 MOZ Operational Deployment (Limited Rollout)

WFP DEEP Machine-Learning Drone Damage Assessment for Post-Cyclone Humanitarian Response (Mozambique)

Social assistance Deep learning