A human-machine collaborative approach measures economic development using satellite imagery

Machine learning using satellite imagery offers accessible ways to infer socioeconomic measures. However, many algorithms need ground-truth data, scarce in many countries. This study presents a human-machine collaborative model predicting grid-level economic development using publicly available satellite imagery and lightweight subjective ranking annotation without ground data. Applied to North Korea and five other least developed Asian countries, the model suggests substantial development in North Korea's capital and state-led projects, highlighting its broad applicability and potential for guiding sustainable development programs.

Donghyun Ahn, Jeasurk Yang, Meeyoung Cha, Hyunjoo Yang, Jihee Kim, Sangyoon Park, Sungwon Han, Eunji Lee, Susang Lee, Sungwon Park