Tue. Sep 26th, 2023
Satlas’ AI Software Enhances Satellite Image Resolution by Fourfold

Geospatial data analysis is set to revolutionize various domains such as agriculture, urban planning, and disaster relief. Microsoft and the Allen Institute for AI (AI2) are actively involved in this revolution through their AI-powered projects.

AI2 has introduced a new software platform called Satlas, designed to explore global geospatial data generated from satellite imagery. The platform starts with images obtained from the European Space Agency’s twin Sentinel-2 satellites. Through deep-learning models, the software analyzes the images to identify features like offshore platforms, solar farms, and tree cover. These features are then displayed on an interactive online map which allows tracking of changes over time.

Training the model for Satlas was a challenging task that involved manually labeling thousands of features in Sentinel-2 imagery. The initial three data products focus on the positions of offshore wind turbines and platforms, the locations of solar farms and onshore wind turbines, and the percentage of tree canopy coverage for 100-square-meter plots of land. The developers are planning to release additional geospatial data products in the future.

In addition to exploring geospatial data, the Satlas team has also developed AI tools to enhance the resolution of satellite imagery. The software uses deep learning models to combine multiple low-resolution images of the same location captured at different times, resulting in a high-resolution image that provides sub-pixel details.

Another project by AI2 called Skylight is examining the marine infrastructure data of Satlas to improve vessel tracking in the sea, especially for identifying illegal fishing activities. The team is actively seeking other use cases for Satlas.

Meanwhile, Microsoft’s AI for Good Lab is working on Project Guacamaya with institutions in Colombia. This project utilizes AI tools to monitor and understand conditions in the Amazon Rainforest, tackling the issue of deforestation. The data from Project Guacamaya, which includes satellite images, hidden cameras, and acoustic sensors, is used to track the health of the rainforest and its species. Colombian conservationists can use this data to raise awareness about specific environmental threats.

Microsoft is also collaborating with organizations in Brazil to deploy AI models for another Amazon-monitoring project called PrevisIA.

With these initiatives, AI is set to revolutionize geospatial data analysis and environmental monitoring, helping to address global challenges.