{
  "id": 560286,
  "title": "Clarification on Synthetic Dataset for Noise2Noise Training",
  "url": "/competitions/czii-cryo-et-object-identification/discussion/560286",
  "author_name": "Success Moses",
  "post_date": "2025-01-30T13:20:00.284000",
  "votes": 1,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Hello,</p>\n<p>I have a basic understanding of cryoET imaging, but I am still getting familiar with the details. I understand that raw tomograms (images) are typically very noisy. To denoise them, a Noise2Noise model is required, but for training such a model, multiple realizations of the same stochastic noise are needed — essentially, the same image captured multiple times with different noise realizations.</p>\n<p>Does the synthetic dataset includes these multiple noisy realizations of the same image, as required for training a Noise2Noise model? I am particularly interested in whether the <a href=\"https://cryoetdataportal.czscience.com/datasets/10441?deposition-id=10310\" target=\"_blank\">dataset</a> provides the necessary conditions for generating paired noisy images for this purpose.</p>",
  "messages": [
    {
      "id": 3110810,
      "postDate": "2025-01-30T13:20:00.283Z",
      "content": "<p>Hello,</p>\n<p>I have a basic understanding of cryoET imaging, but I am still getting familiar with the details. I understand that raw tomograms (images) are typically very noisy. To denoise them, a Noise2Noise model is required, but for training such a model, multiple realizations of the same stochastic noise are needed — essentially, the same image captured multiple times with different noise realizations.</p>\n<p>Does the synthetic dataset includes these multiple noisy realizations of the same image, as required for training a Noise2Noise model? I am particularly interested in whether the <a href=\"https://cryoetdataportal.czscience.com/datasets/10441?deposition-id=10310\" target=\"_blank\">dataset</a> provides the necessary conditions for generating paired noisy images for this purpose.</p>",
      "rawMarkdown": "Hello,\n\nI have a basic understanding of cryoET imaging, but I am still getting familiar with the details. I understand that raw tomograms (images) are typically very noisy. To denoise them, a Noise2Noise model is required, but for training such a model, multiple realizations of the same stochastic noise are needed — essentially, the same image captured multiple times with different noise realizations.\n\nDoes the synthetic dataset includes these multiple noisy realizations of the same image, as required for training a Noise2Noise model? I am particularly interested in whether the [dataset](https://cryoetdataportal.czscience.com/datasets/10441?deposition-id=10310) provides the necessary conditions for generating paired noisy images for this purpose.",
      "votes": 1
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "3110810": "Hello,\n\nI have a basic understanding of cryoET imaging, but I am still getting familiar with the details. I understand that raw tomograms (images) are typically very noisy. To denoise them, a Noise2Noise model is required, but for training such a model, multiple realizations of the same stochastic noise are needed — essentially, the same image captured multiple times with different noise realizations.\n\nDoes the synthetic dataset includes these multiple noisy realizations of the same image, as required for training a Noise2Noise model? I am particularly interested in whether the [dataset](https://cryoetdataportal.czscience.com/datasets/10441?deposition-id=10310) provides the necessary conditions for generating paired noisy images for this purpose."
  }
}