Learning to Recognize Local Sustainable Development from Space

From a mainly growth-focused perspective on development, there is a broad consensus nowadays that development needs to deliver on economic, social, and environmental outcomes.

The main problem is that innovation-driven gains in competitiveness and growth in a capitalist society are in principle indifferent to social and environmental outcomes.

The aim of this project is to contribute to sustainable development by introducing new methods to measure and monitor local economic, environmental, and social outcomes timely, accurately, and cost-efficiently.This work contributes to existing debates on localizing SDGs. However, collecting data locally has been costly, time-consuming, and thus limited in scale.

To overcome several of the limitations in previous research, we suggest a novel approach where a convolutional neural network is trained to recognize and predict locations of sustainable development based on various types of images.

Such an approach would be able to identify emerging areas of sustainable development, identify factors that are associated with progress on environmental and socio-economic outcome indicators, and be possible to implement in monitoring and evaluation.

The result will include maps and tables about local sustainability outcomes for the whole of Sweden and for the last 10-15-years and theory-based explanations for the observed patterns.