Planet Analytic Feeds

Planet Analytic Feeds applies deep-learning and computer-vision models to Planet's near-daily PlanetScope satellite imagery (3.7 m resolution) to detect objects and change automatically across a customer's area of interest, so analysts no longer have to search large volumes of imagery by hand.

The feeds are Road Detection and Building Detection (raster segmentation layers, weekly or monthly), Road and Building Change Detection (vector polygons that flag new construction across country-wide areas), Vessel Detection (oriented bounding boxes with estimated length, width, heading and a confidence score; the Enhanced tier adds vessel class and bunkering status) and Aircraft Detection (aircraft of 20 m or more at rest on airfields). Each subscription covers a chosen area and time period, and can reach back in time through Planet's imagery archive. Planet delivers the feeds off the shelf and can also develop custom models for a customer's analysis.

Results are delivered as GeoJSON and GeoTIFF through the Analytics API, which offers an OGC API - Features interface, and through the Analytic Feeds Viewer, where analysts filter detections by confidence, compare before-and-after imagery, and review and export validated detections. Planet publishes model accuracy for each feed, for example precision 0.89 and recall 0.84 for vessels longer than 25 m in open water.

The New Mexico State Land Office compares road change detections with permit applications and estimates $1M in revenue from the violations it found. Brazil's Policia Federal uses the change detections to watch development in protected forests, and Colombia's Instituto Geográfico Agustín Codazzi (IGAC) added the road and building change detection service in 2025.