The Open Buildings 2.5D Temporal Dataset is helping to fill a major gap in data for population in the developing world. The Google

How we used AI to build our newest Open Buildings dataset

submited by

Style Pass

2024-11-10 09:00:04

The Open Buildings 2.5D Temporal Dataset is helping to fill a major gap in data for population in the developing world.

The Google Research Africa team has released the Open Buildings 2.5D Temporal Dataset, an annual snapshot of building presence, counts, and heights across much of the Global South from 2016 to 2023. This dataset, built using AI and low-resolution satellite imagery, provides critical information for governments, organizations, and researchers working on population growth, urban planning, and resource allocation. The team's innovative use of a teacher-student model setup and multiple satellite image frames enables the extraction of detailed building footprints and heights, making this dataset a valuable tool for understanding and addressing the challenges of rapid urbanization in the developing world.

It uses satellite images to show how buildings change over time in Africa, South and Southeast Asia, Latin America, and the Caribbean.