Satlas, developed by AI2 (Allen Institute for AI), is an advanced AI-powered tool that empowers users to delve into and analyze dynamic Earth changes through meticulously annotated satellite imagery.

This versatile tool relies on monthly geospatial data to uncover transformations across various domains, including marine infrastructure, renewable energy infrastructure, and tree cover. With Satlas, users can investigate specific instances such as offshore wind farms in China and Denmark, deforestation in the Amazon Rainforest, logging activities in Washington State, and the proliferation of wind farms in Scotland.

A standout feature of Satlas is its Super-Resolution capability, where state-of-the-art AI models enhance low-resolution satellite imagery, producing high-resolution global images. This enhancement significantly improves clarity and allows users to visualize specific locations in intricate detail. Satlas harnesses cutting-edge architectures and training algorithms, drawing from the extensive SatlasPretrain dataset, which boasts over 30 terabytes of imagery and an impressive 300 million labels.

Satlas is committed to open accessibility, offering freely available AI-generated geospatial datasets for users to download and analyze offline. Furthermore, it contributes to the research community by releasing AI models and training labels to support ongoing research and development efforts. The dedicated team at the Allen Institute for AI spearheads the development of Satlas, with a mission to advance computer vision capabilities for creating AI systems capable of understanding and reasoning about the world.

In summary, Satlas is a robust tool for exploring and comprehending Earth’s transformative processes through AI-annotated satellite imagery. It serves as a valuable resource for gaining insights into various facets of our planet’s evolution.