Test datasets
Two test datasets have been posted to the data download page. These are unlabeled, but consist of the same six facies categories as the training dataset. Your task is to predict the distribution of these facies using the model you derived from the training dataset. Your predictions should consist of 3D volumes the same size as the test images, with each voxel labeled 1, 2, 3, 4, 5, or 6.

The test datasets are also in SEG-Y format; if you decide to submit your predictions, please do so in the same format with identical headers (the Segyio python utility is useful for this; contact the organizers if you need assistance).

Submission
If you would like your predictions to be (anonymously) included in our discussions during the Workshop, please submit them by Monday, September 28. There are two options for submission:

1) [Preferred]  If you have access to a cloud-based file sharing service (Dropbox, OneDrive, etc.), or a storage account with a Cloud provider (AWS, Azure, GCP, etc.), upload your solutions there and send a download link to halpert@chevron.com.
2) Otherwise, contact halpert@chevron.com to make alternate arrangements.

[Optional] Along with your submission, please provide a short (1-2 sentence) description of your approach and/or any special features of your model. Example: ``2D U-net with input image patches of 500 x 500 pixels, and a loss function based on phase of the Moon??. This will allow the organizers to create a more interesting presentation as we review the strengths and weaknesses of the submissions. 

Thanks again for your participation, and good luck!