Case Studies

MXR-CS-infrastructure growth 12/20 Quickly changing context With high levels of poverty, sustained economic growth, significant investments in infrastructure and extreme vulnerability to climate change—all spread across a logistically challenging landscape of over 7,000 islands—the development context and infrastructure requirements in the Philippines are shiing quickly, oen too quickly to allow decisions based on time- and labor-intensive census data or on-the-ground surveys. High-resolution satellite imagery, in conjunction with machine learning and artificial intelligence, is being used to significantly reduce the time needed to gain an up-to-date understanding of situations on the ground. Whether that's guiding aid in the aermath of a natural disaster or quickly identifying road networks at continental scale, these combined technologies offer unprecedented ability to leverage current data in situations where the context changes frequently. AI development solutions Advances like AI for household wealth prediction, enabled by high-quality, high-resolution satellite imagery, support investment based on the best and most current data and allow for more rapid and effective project completion. That means improved communication, heightened transparency, reduced corruption, increased opportunities and reductions in poverty: for a better world. info@maxar.com maxar.com AI to predict household wealth Imagery to go For a large telecom operator providing a range of products and services, understanding the economic value of housing in an area allows the company to provide the most suitable infrastructure for the needs and likely demand in each location. Additionally, the economic value of housing in an area provides a rapid indication of existing facilities, such as those with receiving or transmission equipment, in a particular region. Our customer, Pacific Data Resources, and a team from Thinking Machines Data Science, Inc. worked together to develop and train a series of AI models to detect houses and extract information from Maxar high-resolution satellite images. Across the Philippines, all residential buildings were tagged and scored by socioeconomic class with high levels of accuracy. Models were able to accurately predict household wealth across the Philippines, completing in months a process that would otherwise have taken many years. Models to accurately detect structures and extract relevant information across the widely varied landscape of the Philippines relied on the use of AI. Thinking Machines Data Science developed the models on their in-house platform, and for this it needed ample and accurate data, in a context of frequent and rapid change (both economic and environmental). Using Maxar satellite imagery on their own platform allowed Niek van Veen's team to create the solution it needed with the high-resolution detail and currency it required. Wealth predictions for specific areas in the Philippines using satellite imagery Sample imagery provided by Maxar

• Cover