Remote Sensing

This page is about the applications of machine learning (ML) in the context of remote sensing. For an overview of remote sensing more generally, please see the Wikipedia page on this topic.

Machine learning can be applied to remote sensing data to infer climate-relevant information. Some selected examples include:

• Mapping power grids and solar panel locations.
• Mapping building footprints.
• Pinpointing occurrences of deforestation.
• Creating an inventory of global greenhouse gas emissions.

Online Courses and Course Materials

• GIS & Remote sensing at Earth Lab, University of Colorado

Community

• EARTHVISION: A workshop regularly held at computer vision conferences. Website here.
• Space and AI: A conference organized by the ESA-CLAIRE AI Special Interest Group on Space. Website here.

Data

Satellite imagery datasets

High-resolution RGB satellite images (for visual predictions)

Public datasets

• United States Geological Survey
• Copernicus (Sentinel satellites)
• NASA Worldview
• Google Earth (permission needed for research use)

Commercial datasets

• DigitalGlobe (up to 31cm resolution)
• Planet (up to 72cm resolution)

Multispectral satellite images (5-13 visible and infrared bands)

Public datasets

• Copernicus (Sentinel satellites)
• BigEarthNet (Sentinel satellites)

Commercial datasets

• Digital Globe
• Planet

Hyperspectral satellite images (up to a few hundred visible and infrared bands)

Public datasets

• United States Geological Survey (Hyperion data)

See also

• Table 1 of the review "Survey of Hyperspectral Earth Observation Applications from Space in the Sentinel-2 Context" for comparisons between sources.

General satellite images

• Awesome Satellite Imagery Datasets: List of aerial and satellite imagery datasets with annotations for computer vision and deep learning.

Street view datasets

• OpenStreetMap