In a groundbreaking initiative to bridge Africa’s infrastructure mapping gaps, Google has leveraged its cutting-edge artificial intelligence (AI) technologies to add millions of previously unmapped buildings across the continent to its Google Maps service. This move is part of a major effort to close digital infrastructure gaps and bring visibility to underserved regions.
Abdoulaye Diack, the Programme Manager for Google Research, revealed that the project uses satellite imagery and machine learning models to detect structures, especially in remote and underserved areas that had never been formally documented before. The goal is to increase transparency, improve urban planning, and aid a variety of industries ranging from emergency response to real estate.
“When we started, many neighbourhoods in Africa weren’t visible in Google Maps. Our project has helped fix that,” Diack told our correspondent. “Almost half of the buildings you now see on Google Maps in Africa came from this initiative.”
The initiative relies heavily on European Sentinel satellite data, which is updated every five days. These images were used to train machine learning models that can accurately detect building footprints. Over time, this effort led to the successful mapping of over 150 million buildings across Africa, encompassing everything from modern cities to informal settlements and rural dwellings.
A key breakthrough in this project was the model’s ability to go beyond simply identifying rooftops. It also estimates building heights by analyzing the length of shadows in the satellite images, factoring in the time of day the image was captured.
“We were able to detect the height of over 1.8 billion buildings worldwide. This wasn’t our original goal, but it has turned out to be one of the most fascinating results of the project,” Diack said. This data has proven crucial for urban planners, emergency responders, and economists who now have better insight into the physical layout of African cities and towns.
The initiative, which began in Ghana, has expanded globally, with AI tools initially developed for Africa now being applied in other parts of the world. The project is particularly remarkable considering the data scarcity often faced in Africa. Diack noted that the lack of high-resolution images pushed his team to innovate by using lower-quality satellite data, which has also enhanced model efficiency.
“We had to be creative and efficient. Our work in Africa, where we faced a lack of high-resolution imagery, forced us to think outside the box,” Diack said. This approach not only solved an immediate need but also paved the way for using AI in a more impactful way globally.
Google’s AI efforts are not limited to mapping buildings. The company’s PlantVillage initiative, which started in collaboration with Makerere University in Uganda, uses smartphone cameras to detect diseases in cassava plants. Diack explained that the AI is capable of identifying plant diseases before they’re visible to the human eye, thereby helping farmers take action before significant damage occurs.
The Google Research team has also been instrumental in building datasets for underrepresented African languages. Diack mentioned partnerships with organizations like Masakhane and various universities to improve speech recognition systems that account for local languages and dialects, allowing AI tools to better serve African communities.
A critical part of Google’s vision for Africa involves AI talent development. Diack highlighted the company’s AI residency program, which trains early-career researchers in collaboration with institutions like the African Institute for Mathematical Sciences. Google is also working with groups like AfricaToML to help software engineers across the continent gain the necessary skills for AI careers, particularly in countries like Nigeria.
“Investing in AI talent across Africa is a key part of what we do. We are working closely with universities and local communities to build capacity and create opportunities for young, talented people,” Diack noted.
For AI to fully thrive in Africa, Diack believes local engagement and data accessibility are essential. He urged that more grassroots efforts are needed to train communities and provide datasets that reflect Africa’s languages and cultural realities.
“We need AI tools that understand the local context to truly serve the people. If they can’t understand you, they can’t help you,” Diack concluded.
Google’s AI-driven mapping project is not just about technological innovation but also about empowering Africa. The initiative offers a unique way to address infrastructure gaps, improve urban planning, and create a more digitally inclusive continent. As AI continues to evolve, it holds the promise of further revolutionizing industries across Africa, from agriculture to urban planning and beyond.