Pretrained AI Models

Ready-to-Use Geospatial Deep Learning Models available in the ArcGIS Living Atlas of the World.

This story map walks you through several examples of how  Esri's pretrained models  can be used to extract features, classify land cover or detect objects in imagery, point clouds or even video. To explore all the pretrained models, check out  Living Atlas . Some of these are shown below:


Feature Extraction

Building Footprint Extraction (USA)

Imagery: High resolution satellite imagery Geographies: United States

Powered by Esri
Powered by Esri

Building Footprint Extraction (Africa)

Imagery: High resolution satellite imagery Geographies: Africa

Powered by Esri
Powered by Esri

Road Extraction (North America)

Imagery: High resolution satellite imagery Geographies: United States, Canada

Powered by Esri
Powered by Esri

Swimming Pool Detection

Imagery: High resolution satellite imagery Geographies: United States

Car Detection

Imagery: Drone imagery (5-20 cm) Geographies: United States

Solar Panel Detection

Imagery: Drone imagery (5-15 cm) Geographies: United States

Shipwreck detection

Imagery: Bathymetric Attribute Grid (BAG) Geographies: Global

Ship Detection

Imagery: Synthetic Aperture Radar (SAR) Geographies: Global


Land Cover Classification

Land Cover Classification (Sentinel 2)

Imagery: Sentinel-2 (10m resolution) Geographies: Europe and United States

Land Cover Classification (Landsat 8)

Imagery: Landsat-8 (30m resolution) Geographies: United States

Human Settlements (Landsat 8)

Imagery: Landsat-8 (30m resolution) Geographies: United States

Human Settlements (Sentinel 2)

Imagery: Sentinel-2 (10m resolution) Geographies: Europe and United States

Point Cloud Classification

Power Line and Tree Point Classification

Input data: Unclassified Point Clouds Geographies: Global

Image Redaction

Face and License Plate Blurring

Object Tracking