AI Assesses Infrastructure Quality in Africa

Monitoring infrastructure quality in developing regions is a major goal for international aid organizations, unfortunately gathering up to date and reliable data is expensive and a time-consuming process. To alleviate the burden on aid organizations, researchers from Stanford University developed a deep learning-based method that can assess infrastructure quality from satellite imagery.
“Basic infrastructure availability in developing regions is a crucial indicator of quality of life. Reliable infrastructure measurements create opportunities for effective planning and distribution of resources, as well as guiding policy decisions on the basis of improving the returns of infrastructure investments,” the researchers stated in their paper. “Currently, the most reliable infrastructure data in the developing world comes from field surveys, and these surveys are expensive and logistically challenging.”
Using NVIDIA Tesla GPUs on the Google Cloud, with the cuDNN-accelerated TensorFlow deep learning framework, the team trained their neural network on 10m and 30m resolution satellite imagery from the Sentinel 1 and Landsat 8 satellites. The satellites provided the team with thousands of images encompassing 36 countries in Africa.
“Our work provides the ability to assess infrastructure in an accurate and automated manner, to supplement the spatial extent of field survey data, and to generate predictions in unseen regions. To the best of our knowledge, we are the first to use CNNs with Sentinel 1 imagery for social development research,” the team said.