SATLAS, developed by the Allen Institute, allows users to analyze geospatial data using super-resolution imagery, enhancing low-resolution satellite images with advanced computer vision algorithms.

Features:

• AI Models for Remote Sensing: Discover AI models trained on extensive remote sensing datasets, including satellite imagery and labels from the Satlaspretrain dataset.
• Access to Geospatial Datasets: Provides access to AI-generated geospatial datasets for download and offline analysis.
• Objective of SATLAS Team: Aims to advance computer vision boundaries and develop AI systems capable of perceiving, exploring, learning, and reasoning in real-world scenarios.

free

Is it possible to download the data?
You can download all the data shown in Satlas, as well as the training data and model weights, for offline analysis from this link. Information can be accessed through monthly maps beginning in January 2016. Marine infrastructure and renewable energy infrastructure data are released monthly in GeoJSON format, while tree cover data is provided monthly in GeoTIFF format.
How precise is the data?
Our AI models produce data with exceptional accuracy. Nevertheless, AI systems are never flawless and various factors tend to diminish their performance. Our Data Validation Report provides an in-depth analysis of the accuracy of each Satlas geospatial data product.Why are satellite images unavailable for my specific time and location?
The satellite imagery in Satlas is sourced from the Sentinel-2 satellites, and they do not provide complete coverage of the entire planet during each pass. If satellite imagery is unavailable for your specific time and location, consider selecting a neighboring month to access an image from a different orbit.