Now you can contribute to space data research and help us accelerate the development of Swedish AI by participating in our Space Data Hackaton on the theme cloud occluded satellites images. Participants will get access to Copernicus data, support from experts and the option to follow topic specific webinars/workshops from experienced mentors, before and during the hackathon.
And of course there is a jury and a grand prize!
Why hack on satellite images?
The availability of large amounts of satellite imagery data through the European Copernicus project and open source platforms like the OpenDataCube (ODC) greatly bolsters the opportunities to apply classical machine learning and deep learning algorithms.
These models can be employed to meet many challenges. One challenge often arising in the analysis of satellite data is that of cloud occlusion. Depending on the time of the year and the location of the Area of Interest (AoI) clouds can fully or partially cover the image and hamper the analysis. There are, however, many methods available for the estimation of missing data.
Two parts of the challenge
Part 1 in this challenge is to use the Swedish satellite data (provided through ODC) to train a model that is able to estimate certain statistics for an occluded area based on the information from the surrounding area, as reliably and accurately as possible.
Part 2 in this challenge is to develop a viable business case for the solution developed or from techniques utillsed to develop the solution. Team will be given a framework to follow for this aspect of the challenge.
The aim of the Space Data Lab is to increase the use of space data to progress our society, to develop our industry and to make the Earth a better place. It will serve as a national resource for the Swedish authorities in their work on Earth observation data and the development of AI-based space data analysis. The objective is to establish the data, technology and methods that will be required towards the systematic development of services and applications that are based on space data, here at the Space Data Lab.