Somewhere on your street, a satellite passed overhead. It captured your roof, your neighbor’s garage, and that garden shed you built last summer. Now all of it sits inside a digital atlas alongside 2.75 billion other buildings scattered across every continent.
Join a community of 14,000,000+ Seekers!
Subscribe to unlock exclusive insights, wisdom, and transformational tools to elevate your consciousness. Get early access to new content, special offers, and more!
A research team in Germany has done something no one has accomplished before. They counted and modeled nearly every human-made structure on the planet. Not just skyscrapers in Manhattan or apartments in Tokyo. Every farmhouse in rural France. Every market stall in Lagos. Every fishing hut along the Mekong River.
You can zoom in right now and find your own home rendered in three dimensions. But what makes this map truly remarkable has little to do with satisfying curiosity about your own address. It has everything to do with what 2.75 billion buildings reveal about how we live, where we struggle, and what we might become.
A Billion More Buildings Than Anyone Else Has Counted
Professor Xiaoxiang Zhu leads the Chair of Data Science in Earth Observation at the Technical University of Munich. Her team spent years building what they call the GlobalBuildingAtlas, and it dwarfs every previous attempt to catalog human structures.
Before this project, the largest global dataset contained roughly 1.7 billion buildings. Zhu’s team added over one billion more. They captured regions that earlier maps had ignored or underestimated, including vast stretches of Africa, South America, and rural communities worldwide.
Of the 2.75 billion structures in the atlas, 97 percent appear as simplified 3D models. You can see not just where a building sits but how tall it stands and how much space it occupies. At a resolution of 3 by 3 meters, the map is 30 times finer than comparable databases. It picks up shapes and sizes that previous efforts missed entirely.
Anyone can access the atlas online at no cost. Within days of launch, nearly 280,000 people visited the site, causing temporary loading issues as curious users searched for their homes, their hometowns, and places they had only read about.
How Satellites and Machine Learning Built a Planet-Wide Census
Creating such a map required an extraordinary amount of raw material. Zhu’s team collected more than 800,000 satellite images from 2019, all captured by PlanetScope, a constellation of Earth-observation satellites.
But images alone were not enough. Distinguishing a building from a road, a tree, or a rocky outcrop demanded machine learning algorithms trained on thousands of examples. Researchers fed the system labeled data from cities across Europe, North America, and parts of Asia. Over time, the AI learned to recognize rooftops, estimate heights, and separate one structure from another.
Height estimation proved especially challenging. In wealthy nations, governments have flown planes equipped with LiDAR sensors that measure elevation with laser precision. Those readings helped train the AI. But such data barely exists for Africa, and coverage remains patchy across much of South America and Southeast Asia. Researchers had to teach the system to make educated guesses based on shadows, textures, and patterns it had learned elsewhere.
To fill the remaining gaps, the team fused their AI-generated results with existing datasets from OpenStreetMap, Google, and Microsoft. A quality-guided strategy ensured that whichever source offered better accuracy in a given region would take priority. By combining multiple inputs, Zhu’s team produced a map far more complete than any single source could deliver.
Why 3D Models Beat Flat Maps
Traditional maps show buildings as flat shapes. You see where a structure sits, but you learn nothing about whether it rises two stories or twenty. For many purposes, that limitation matters.
Consider two neighborhoods with identical footprints on a 2D map. One is a crowded, informal settlement where families squeeze into single-story homes with shared walls. Another is a planned urban district where residents live in spacious high-rise apartments. Flat maps treat them the same. A 3D model captures the difference.
“3D building information provides a much more accurate picture of urbanization and poverty than traditional 2D maps,” Zhu explained. “With 3D models, we see not only the footprint but also the volume of each building, enabling far more precise insights into living conditions.”
Her team introduced a new metric they call building volume per capita. It measures the total building mass in a region relative to the number of people living there. Wealthier areas tend to show higher volumes per person, with larger homes, taller structures, and more generous spacing. Poorer areas often display the opposite.
When researchers compared this metric against GDP per capita, the correlation was striking. Building volume per capita aligned with economic development more closely than traditional area-based measures. It offers a new lens for spotting inequality and tracking progress.
What 2.75 Billion Buildings Reveal About Global Inequality
Numbers on a spreadsheet rarely stir emotion. But mapped across the globe, they tell a story impossible to ignore.
Asia leads every category. With 1.22 billion buildings and a total volume of 1.272 trillion cubic meters, it reflects both an enormous population and rapid urban growth in places like China, India, and Southeast Asia.
Africa presents a striking contrast. It’s 540 million buildings that outnumber Europe’s 403 million. Yet total building volume in Africa reaches only 117 billion cubic meters, far below Europe’s 763 billion. African buildings tend to be smaller and shorter, a pattern consistent with widespread informal construction and limited resources.
At the country level, disparities grow even starker. Finland boasts the highest building volume per capita on Earth. Niger, in West Africa, has just 0.36 percent of Finland’s figure. A person in Niger lives surrounded by 27 times less building volume than the global average.
Such gaps are not merely academic. They signal where housing shortages bite hardest, where infrastructure lags behind population growth, and where governments must direct resources if they hope to meet basic needs.
From Disaster Response to Climate Planning
Zhu hopes the atlas will do more than illuminate inequality. She envisions it as a tool for action. Urban planners can use the data to identify neighborhoods that need new schools, hospitals, or transit lines. Emergency responders can assess which buildings sit in flood zones or earthquake-prone areas. Climate researchers can sharpen models of energy demand and carbon emissions, since buildings account for nearly 40 percent of global CO₂ output.
Germany’s Aerospace Center has already begun exploring the atlas for disaster response. Under an international charter, space agencies share satellite data during emergencies like hurricanes, earthquakes, and wildfires. Having a pre-existing 3D model of affected areas could speed damage assessments and guide rescue efforts.
“We introduce a new global indicator: building volume per capita, the total building mass relative to population – a measure of housing and infrastructure that reveals social and economic disparities,” Zhu noted. “This indicator supports sustainable urban development and helps cities become more inclusive and resilient.”
By 2050, nearly seven in ten people will live in cities. Managing that growth wisely will require better data than we have ever had. Zhu’s atlas offers a starting point.
Where Gaps and Errors Remain
No dataset of this scale arrives without flaws. Zhu’s team is candid about limitations. Africa remains the weakest link. Without local LiDAR measurements to train the AI, height estimates in that region carry greater uncertainty. Models trained on European and North American buildings may not generalize perfectly to informal settlements or traditional architecture found across the continent.
High-rise buildings also pose challenges. Across South America and parts of Asia, the AI tends to underestimate how tall skyscrapers actually stand. South America shows the highest height estimation errors, with a root mean square error of 8.9 meters.
A small disclaimer on the interactive map warns users to expect imperfections. “This is a machine-learning-derived product. It may contain errors,” it reads.
Zhu’s team plans to improve accuracy over time. Future versions may incorporate temporal data, showing how cities expand or contract year by year. Such time-series analysis could reveal patterns invisible in a single snapshot.
Nearly 280,000 Visitors in Days
Interest in the atlas has been immense. Within days of its public release, nearly 280,000 people attempted to access the interactive map. Traffic overwhelmed servers, leaving some visitors unable to load certain regions.
Most seemed driven by simple curiosity. They wanted to find their own homes, see how their neighborhoods looked from above, and compare their cities to places halfway around the world. But researchers, urban planners, and aid organizations also took notice, recognizing the atlas as a resource unlike anything previously available.
All data and code sit on GitHub under an open license. Anyone can download the information, run their own analyses, or build new tools on top of what Zhu’s team created. It is a model of open science applied to one of humanity’s most basic questions. Where have we built, and what does it look like?
What a Map of Every Building Says About Us
Seeing 2.75 billion structures laid out across continents does something unexpected. It makes humanity’s presence feel both immense and fragile. We have built homes, hospitals, factories, and temples on nearly every patch of habitable land. Yet from orbit, each building is just a tiny cube among billions.
For most of human history, no one could have imagined counting every structure on Earth. Now a team of researchers has done exactly that, using satellites and algorithms to catalog what centuries of human effort have produced. It reminds us that our species does not simply inhabit the planet. We reshape it, one building at a time.
Looking at the atlas, you can trace patterns of migration, wealth, and ambition. Dense clusters appear in East Asia. Sprawling suburbs stretch across North America. Sparse settlements dot the Sahel. Each marker on the map represents someone’s home, workplace, or place of worship. Together, they form a record of where we chose to put down roots and why.
Perhaps the most striking lesson is how unevenly we have built. A person in Finland lives surrounded by far more building volume than someone in Niger. Such disparities are not new, but seeing them rendered in 3D makes them harder to ignore. Data like this can push governments and organizations to ask harder questions about who has shelter, who lacks it, and what we can do about it.
“Buildings anchor human life and define the form and function of urban environments,” Zhu’s team wrote in their published paper. “3D insights are essential for urban planning, infrastructure management and policy-making – especially in resource-limited contexts where the strategic allocation of funding and intervention is critical.”
Maps have always helped us understand our place in the world. Ancient cartographers drew coastlines and trade routes. Modern satellites capture ice sheets and deforestation. Now we have a tool that shows humanity’s physical footprint in three dimensions. It is both a mirror and a challenge, reflecting what we have built while asking what we might build next.
22
0
0
0
