Everything we share is sourced from publicly available data served online, APRA public-records requests, government and historical archives, open datasets and Git repositories, and measurements we take ourselves.
Some of what you see, we measure ourselves, in the open. We fuse Rhode Island's three-inch aerial imagery with USGS LiDAR to read building heights, roofs, tree canopy, and street furniture, and we trace the footprints the public data misses. The heavy lifting runs on open-source engines, OpenCV and scikit-image for the vision work, PDAL and laspy for the point clouds, and Meta's Segment Anything for the tricky outlines, alongside Apple's on-device Core ML and Vision framework. It is all computed locally on our own Macs, no cloud and no black box. The inputs are the public datasets listed above, and the exact method is written up on our methodology page.
No sources match that filter.