Multi-Task Deep Learning for Predicting Poverty From Satellite Images

Abstract

Estimating economic and developmental parameters such as poverty levels of a region from satellite imagery is a challenging problem that has many applications. We propose a two step approach to predict poverty in a rural region from satellite imagery. First, we engineer a multi-task fully convolutional deep network for simultaneously predicting the material of roof, source of lighting and source of drinking water from satellite images. Second, we use the predicted developmental statistics to estimate poverty. Using full-size satellite imagery as input, and without pre-trained weights, our models are able to learn meaningful features including roads, water bodies and farm lands, and achieve a performance that is close to the optimum. In addition to speeding up the training process, the multi-task fully convolutional model is able to discern task specific and independent feature representations.

Predicting socioeconomic indicators using transfer learning on imagery data: an application in Brazil

“…Many studies have analysed the interaction between satellite imagery and income and poverty data. Approaches differ regarding the type of satellite imagery: only nighttime lights (Elvidge et al 2009;Yu et al 2015); only daytime images (Engstrom et al 2017a, b;Pandey et al 2018;Bai et al 2020); a combination of both (Xie et al 2016;Jean et al 2016;Perez et al 2017;Zhao et al 2019).…”

“…The linear regression results indicate that poverty can be partially explained by spatial and spectral features, since the R 2 for each estimated poverty level was slightly higher than 0.5. Focusing on the predictive power of daytime imagery regarding poverty in India, Pandey et al (2018) obtained 1920 × 1920 sized images from Google Static Maps API. According to the authors, this was the first study to report deep learning experiments on images of such size.…”

“…In general this is not the case when predicting socioeconomic variables, where data is scarce. Transfer learning is a possible method for dealing with this issue and it was used before by Xie et al (2016), Jean et al (2016), Perez et al (2017) and Pandey et al (2018) to predict socioeconomic indicators.…”

“…Dealing with these issues is crucial and requires accurate and updated data for research and effective public policy design. However, as stated by Zhao et al (2019), Pandey et al (2018), Engstrom et al (2017a) and Jean et al (2016), censuses and other ground level surveys are expensive, time consuming Abstract Censuses and other surveys responsible for gathering socioeconomic data are expensive and time consuming. For this reason, in poor and developing countries there often is a long gap between these surveys, which hinders the appropriate formulation of public policies as well as the development of researches.…”

Using Satellite Data to Guide Urban Poverty Reduction

“…Object identification of buildings and their quality as well as cars, combined with geospatial data on roads and farmland, are used to predict municipal poverty rates in Sri Lanka (Engstrom et al ., 2017). Similarly, object identification of water source, roof quality and lighting source are used to predict poverty at sub‐district levels in Uttar Pradesh, India (Pandey et al ., 2018). Identification of roof quality, but not poverty itself, also provides the basis of allocation of anti‐poverty transfers in some villages in Uganda and Tanzania (Abelson et al ., 2014).…”