The Challenge
Your task is to denoise “shabby” documents, reversing the effects of Augraphy. There are several methods to accomplish this, so feel free to experiment and find which is most effective.
Background
The Augraphy Project is using this Kaggle competition as a way to validate and tune the experimental Shabby Pages dataset. The competition data is not the final Shabby Pages data, and represents only a subset of the whole. We will communicate the official release of the full version of the Shabby Pages dataset at a later time. For now, the uncropped full version of the competition data is available here.
Rules
- All submissions must be written in Python, but any library can be used.
- No online solutions-as-a-service may be used.
- Groups of up to 8 people are allowed.
Important Dates
The competition will open on March 21st and will begin once there are 10 submissions or 20 competitors, whichever comes first.
The competition will run for one month from the date it opens, which we will announce here, on the competition page, and in the Slack. Good luck!