NASA Harvest Field Boundary Detection Challenge
Can you detect field boundaries for Rwandan smallholder farmers in a satellite image dataset?
Prize
$5 000 USD
Time
~1 month to go
Participants
126 active · 568 enrolled
Helping
Rwanda
Winning solution requirements
Connect · 12 Jan 2023, 09:30 · 3
Dear all,
When the competition closes we will request the top 10 solutions. We will go through these solutions until we find the top 3 that run, reproduce the same score and are useful to the host.
Once we have determined the top 3 solutions you will be required to work with Radiant Earth to publish your solutions on ML Hub.
You will be required to use this MLHub model template repository to provide model documentation and source code.
https://github.com/radiantearth/mlhub_model_template
Once your solution is on MLHub payment will be made.
I look forward to seeing your final solutions!
All the best,
What does "useful to the host mean"? Is there a metric you are using to determine that?
Useful to the host means following the rules/ ensuring that your solution can be used outside of this competition context.
Most importantly.
The solution must use publicly-available, open-source packages only.
You may use only the datasets provided for this competition. Automated machine learning tools such as automl are not permitted.
If the challenge is a computer vision challenge, image metadata (Image size, aspect ratio, pixel count, etc) may not be used in your submission. You are not allowed to use the latitude and longitude of the pixels as predictors in your model either.
You may use pretrained models as long as they are openly available to everyone.
An important part to ensuring your code is published on MLhub and that you receive your prize is inference portion of your script!
Read this template, paying attention to "Focus on inferencing" https://github.com/radiantearth/mlhub_model_template