{
  "fetched_at": "2026-10-19T09:37:40Z",
  "source_url": "https://www.kaggle.com/competitions/denoising-shabby-pages/data",
  "pages": [
    {
      "name": "data-description",
      "content": "You are provided 2 sets of images, train, and test. Each of the images in those two folders has a clean and shabby version, making it possible to train and test your own denoising model. These images are of documents synthetically noised with the [Augraphy](https://github.com/sparkfish/augraphy) library, which simulates real-world paper printing, faxing, scanning, and copy machine processes. Your task is to create an algorithm to remove the noise from images in the test set.\n\n**Note:** You **MUST** use make-submission.py to produce your submission file.\n\n## Files\n\n*   **train/train_shabby** - The training data ran through the Augraphy pipeline.\n*   **train/train_clean** - The training data ground truth.\n*   **validate/validate_shabby** - The validation data ran through the Augraphy pipeline.\n*   **validate/validate_cleaned** - The validation data ground truth.\n*   **test/test_shabby** - The test data ran through the Augraphy pipeline.\n*   **test/test_cleaned** - The test data ground truth.\n*   **competition_images.zip** - The shabby versions of the solutions. Denoise these documents, run your solution through make-submission.py, and submit the CSV file.\n*   **sample_submission.csv** - A sample submission file in the correct format.\n*  **make-submission.py** - Change the cleaned_images_dir path in this file and run it to generate your submission file.",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Description",
      "content": "# **Your Challenge**\nYour task is to denoise “shabby” documents, reversing the effects imposed on them by [Augraphy](https://github.com/sparkfish/augraphy). There are several paths one might take to accomplish this goal, so feel free to experiment and find which leads to doom and which leads to glory.\n\n![](https://miro.medium.com/max/1400/1*-Gif4QqHRoUotVE_AVvv6A.png)\n### The task is simple: denoise shabby documents, like the right image, to make them look as clean as possible, like the left image.\n\n# **Historical Significance**\nDocuments are exposed to harsh conditions once scribed and enter the world. Most of human history is captured in paper form, some going back thousands of years -- yet the knowledge they hold often degrades very quickly. Even in the modern world, as those documents are scanned, faxed, copied, and printed, there are many distortions that get introduced that obscure their original form. Those distortions, or \"noise\", prevent them from being captured in a more immortal, digital form. That is the problem we intend to solve, with your help.\n\nDepart with us on the hero's journey, to chart the uncharted path that restores luster to that which was once lost.\n\n# **Due Date**\nThe competition is set to end based on the accuracy of the top denoising model. The winner of the competition will be the first person with a denoising model with an RMSE value of less than 0.05. To train models with this accuracy, look at how you can create more data under the Data tab.\n\n# **Rules**\nAll code submissions must be written in Python, but any library can be used\nNo online solutions-as-a-service may be used\nGroups of up to 3 people are allowed\n\n# **Augraphy Workshop**\nUnlike most competitions, this one will feature a workshop that describes how Augraphy works and how you can use it to supercharge your submissions to improve your odds of winning. Once the competition receives more interest, we'll poll competitors on Slack and determine a date and time for the workshop. Check back here or in the Slack channel for the announcement!\n\n# **Communication**\nAll official communication will be done through Slack.\n[Join in the discussion](https://join.slack.com/t/denoisingshabby/shared_invite/zt-160i3x0c7-3mGB81~5M0mt2t0zcq2N_Q) happening right now on Slack to see how others are approaching a solution and to access other opportunities to learn!\n\n# **Prizes**\nThe grand prize offered by our sponsors are [detailed separately…](https://denoising-shabby.com/prizes/)\nWe are awarding the author of the best write-up and code documentation with the chance to meet and collaborate with the Augraphy Team on the Augraphy paper, which will be submitted to global AI conferences.\nWe are also awarding 20 Augraphy T-shirts! 10 for the highest-scoring submissions on the leaderboard, and 10 for the most active participants in the Kaggle competition discussion.\n\n# **Website**\nAll of this information and more is accessible at [https://denoising-shabby.com/](https://denoising-shabby.com/).\n\n# **Shabby Pages**\nThe full Shabby Pages dataset is under active development, and you can help by participating in this competition! We'll release the full set later, but for now, you can grab the cropped set available on the Data page of this competition, or you can grab that and the full-size set from [here](https://drive.google.com/drive/folders/12TtuZHgtTeFyJEhV3DECBXYFlYt4tGUC?usp=sharing).",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Evaluation",
      "content": "Submissions are evaluated on the root mean squared error between the pixel intensities of the cleaned images and the pixel intensities of the original images, where both sets of images have been grayscaled.\n\n# Submission File\nForm the submission file by grayscaling your cleaned images, assigning each image an ID number, and producing a row per pixel in each image, in the following format:\n\n```\n1_1_2,0\n```\n\ncorresponding to the following data:\n1. which picture the data belongs to (filename)\n2. the pixel's row image\n3. the pixel's column\n4. the intensity of the pixel (0, black, to 1, white).\n\n\nThe file should contain a header and have the following format:\n```\nid,value\n1_1_1,1\n1_2_1,1\n1_3_1,1\n```\netc.\n\nA sample submission file is provided in the competition data.",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "rules",
      "content": "## **Python**\nAll submissions must be in Python. However, any library may be used.\n\n## **Online Solutions**\nNo online solutions-as-a-service may be used. Be original and creative! We want to see your solutions!\n\n## **No private sharing outside teams**\nPrivately sharing code or data outside of teams is not permitted. It's okay to share code if made available to all participants on the forums.\n\n## **Team Limits**\nThere is a maximum team size of 3 people. However, there will only be one prize per team. Therefore, we encourage competing alone.\n\n## **Submission Limits**\nYou may submit a maximum of 5 entries per day.\n\n# **Have Fun and Good Luck!**",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Augraphy",
      "content": "# **The Augraphy Project**\n\n\nAugraphy is a Python library that creates multiple copies of original documents through an augmentation pipeline that randomly distorts each copy – degrading the clean version into dirty and realistic copies rendered through synthetic paper printing, faxing, scanning, and copy machine processes.\n\nThe dataset created for the Denoising Shabby Pages competition was generated using Augraphy pipelines to create realistic old and noisy documents from “born digital” sources. This simulation of realistic paper-oriented process distortions creates large amounts of training data for AI/ML processes to learn how to remove those distortions.\n\nFork and star [Augraphy](https://github.com/sparkfish/augraphy) on GitHub to help the team and improve the project!\n\n[![GitHub forks](https://img.shields.io/github/forks/sparkfish/augraphy.svg?style=social&label=Fork&maxAge=2592000)](https://GitHub.com/sparkfish/augraphy/network/)\n[![GitHub stars](https://img.shields.io/github/stars/sparkfish/augraphy.svg?style=social&label=Star&maxAge=2592000)](https://GitHub.com/sparkfish/augraphy)\n\n![The Augraphy Project](https://raw.githubusercontent.com/sparkfish/augraphy/dev/images/logo/augraphy.png)\n\nThe ShabbyPages dataset is compiled through the Augraphy pipeline. If you would like to run your own documents through the pipeline to create your own training/testing data, we encourage you to do so! [You can access the pipeline here.](https://github.com/sparkfish/shabby-pages/blob/dev/shabbypipeline.py)\n\n[You can use these clean documents to run through the pipeline.](https://drive.google.com/drive/folders/1NJhrxo5wyQnyPUKuRkIMOmcW7Zkd2J_0?usp=sharing)",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    }
  ]
}