Important: Read the terms and conditions before agreeing to participate.
This year’s data challenge will focus on the prediction of visible imagery at night using thermal infra-red (IR) observations. The overall ambition is to generate predicted visible imagery with sufficient quality such that existing data products that rely on daytime data can be extended to 24 hours.
Go to the tab "Starting kit" to download the data needed for this competition.
Please use the forum if you have any questions
Good luck to everyone!
> Recording of the introduction: https://zoom.us/rec/share/acNtcnJqdwHHjYs1dYfPp4X18bfsZMLLl1_Mx8yjK0dNj6Mey95SWFmGCqja9lx9.ccxYnjvxU2UUKZcu Passcode: s.Qfx1?Y
> Slides: https://docs.google.com/presentation/d/1Cbh5fFsd4yUcFs7DBeemqyOd7k4Ph4n6BgYjW_PMg-Y/edit?usp=sharing
> Data source: https://registry.opendata.aws/noaa-goes/, we also used Satpy for processing https://doi.org/10.5281/zenodo.596324
Every submission will be ranked according to its mean SSIM index on the test set according to the competition phase. During the development phase, your will see your score on half of the test set. At the second phase, you'll have one only one submission to do and the score on the second half of the test set will be kept hidden until the Hackathon wrap up on Friday.
By participating you agree:
Download | Size (mb) | Phase |
---|---|---|
Public Data | 9.925 | #1 Validation |
Public Data | 10.526 | #2 Test |
Start: Sept. 30, 2020, 2 a.m.
Description: Development phase: Submit your prediction on the validation set.
Start: Sept. 30, 2020, 10:30 a.m.
Description: Test phase: Submit your prediction on the test set.
Never
You must be logged in to participate in competitions.
Sign In# | Username | Score |
---|---|---|
1 | dfulu | 0.86 |
2 | Shahine | 0.77 |
3 | arisml | 0.52 |