Keywords: deep learning, transfer learning, poverty mapping
TL;DR: We propose applying convolutional neural networks to vegetation index to perform poverty prediction.
Abstract: Accurate and timely estimates of economic status are critical to policy-makers in the world's poorest countries. Previous work has applied convolutional neural networks (CNNs) to high-resolution satellite imagery to perform poverty prediction. Although promising, such imagery has limited access and lacks a temporal signal. We show that publicly available, moderate-resolution vegetation index can be used with CNNs to produce equally accurate poverty estimates for developing countries that are heavily dependent on agriculture. In contrast to previous work, the continuous streaming of well known vegetation indices also allows us to update our estimates in light of weather shocks, opening up the possibility of making dynamic poverty mapping at minimal cost.