SpaceNet Data Now Available in GroundWork

SpaceNet Data Now Available in GroundWork

Incorporate high-resolution satellite imagery into your labeling projects for free

Today, Azavea is announcing the launch of a new partnership with CosmiQ Works, which will allow anyone to access their corpus of over 100GBs of urban, high-resolution satellite imagery inside of GroundWork. All users can now easily incorporate this openly licensed imagery into their own labeling projects for free.

screengrab of spacenet data in the groundwork app
SpaceNet data is now available directly within GroundWork

A leader in satellite imagery analysis

SpaceNet is a familiar fixture for many in the earth observation industry–since its founding in 2016, it has grown to become the premier data science competition focused on satellite imagery analysis. Having recently announced the winners of the sixth SpaceNet competition, CosmiQ Works crossed $250,000 in total prizes awarded to date.

Beyond the tremendous value that SpaceNet brings to the earth observation community by attracting machine learning engineers to familiarize themselves with the quirks of geospatial data in droves, the accompanying labeled datasets that they have released are unrivaled in terms of scale and precision. In the past, if a company or researcher were interested in testing out computer vision techniques on satellite imagery, there was no such data readily available–SpaceNet has more than filled that gap. And CosmiQ Works, more generally, continues to publish state of the art research and openly licensed software on topics ranging from super-resolution, to synthetic data, to repeatable ML workflows

Gain new insights about urban areas

Past SpaceNet competitions have focused exclusively on the identification of roads and buildings, two of the highest-value cartographic features visible in satellite imagery. But urban satellite imagery contains vast amounts of information beyond that–you can use these data to estimate tree canopy, calculate impervious surface, count vehicles, inventory solar panel capacity, and much more. By making SpaceNet imagery available in GroundWork and easily accessible for anyone to label without restriction, we hope to enable creative uses for this imagery whether motivated by commercial prospects, research initiatives, or just plain curiosity.

Screenshot of Raster Vision application.
Raster Foundry

We’re helping organizations extract insights with machine learning.

Learn more

To explore this imagery, you can sign up for a free account here and get started. And if you’re interested in competing in future SpaceNet competitions, be sure to also stay tuned to CosmiQ Works’ blog for an impending announcement about SpaceNet 7.