Assigning a Grade: Accurate Measurement of Road Quality Using Satellite Imagery
Roads are critically important infrastructure to societal and economic development, with huge investments made by governments every year. However, methods for monitoring those investments tend to be time-consuming, laborious, and expensive, placing them out of reach for many developing regions. In this work, we develop a model for monitoring the quality of road infrastructure using satellite imagery. For this task, we employ a unique dataset of road quality information on 7000km of roads in Kenya combined with 50cm resolution satellite imagery. We create convolutional neural network-based models for a binary classification task as well as a comprehensive 5-category classification task with both a standard train-test split as well as a more challenging held-out scenario (i.e., for never-before-seen roads). We believe that these preliminary results are well-positioned for impact on a broad set of transport applications in developing regions.
This is joint work with Gabriel Cadamuro (at the University of Washington) and Aggrey Muhebwa (at UMass).
Jay Taneja is an Assistant Professor of Electrical and Computer Engineering at the University of Massachusetts, Amherst. He develops and studies applications of sensing and communications technology on the measurement and management of infrastructure systems in developing regions. Prior to joining UMass, he was a Research Scientist leading the Energy team at the IBM Research - Africa lab in Nairobi, Kenya, from 2013 to 2016. There, he focused on developing technology to improve electricity reliability and access in sub-Saharan Africa, collaborating with utilities and other energy service companies. He earned his Ph.D. and M.S. in Computer Science at the University of California - Berkeley, where for his dissertation work, he built and studied supply-following electricity loads that change electricity consumption to match fluctuations of increasingly renewable electricity supplies.