Get Started

GeoVisual Search works on two different imagery catalogs, showcasing multiple resolutions as well as national- and global-scale search capabilities. Explore these examples below.

Center pivot irrigation

Airport runways

Wind turbines

How GeoVisual Search works

Tile source maps

We divide the globe into overlapping grids of tiles—and then use these grids to capture features that would otherwise run across tiles.

Extract features

For each image tile, we use a deep neural network to extract features such as shapes, colors and texture. These features can include the visible and non-visible spectrum.

Query features

Given a query image, we can calculate a “visual distance” between the query features and the features extracted from each image in the comparison set.

Match results

We match the tiles with the smallest “visual distance” to the query tile.

Search the globe

Descartes Labs created two base layer map composites for GeoVisual Search utilizing Landsat 8 for global coverage and NAIP for the US.

Aerial Imagery (NAIP)

Resolution: 1 meter per pixel

Coverage: United States

Number of pixels processed: 31.5 trillion

Number of tiles processed: 1.9 billion

Landsat 8

Resolution: 20 meters per pixel

Coverage: Global

Number of pixels processed: 3.4 trillion

Number of tiles processed: 205 million