Combining satellite imagery and machine learning to target social protection in Pakistan
UC Berkeley, Center for Effective Global Action
Ref: SRC-002-PAK-001
Accessed: 3/23/2026
Summary
CEGA project page describing the research initiative to develop geographic poverty targeting using ensemble deep learning for Pakistan's Sindh Province. Confirms the project was initiated in response to twin shocks of Covid-19 and 2020 monsoon flooding. Documents the Social Protection Strategic Unit's role in commissioning the research and the three-stage validation framework. Notes the system predicts poverty in grid areas ten times smaller than earlier models with lower exclusion errors.
View Harvard reference
Center for Effective Global Action (CEGA) (2023) 'Combining satellite imagery and machine learning to target social protection in Pakistan'. Berkeley: UC Berkeley. Available at: https://cega.berkeley.edu/collection/ml-targeting-sindhpakistan/ (Accessed: 23 March 2026).