{
  "id": 216572,
  "title": "A little bit about augmentations",
  "url": "/competitions/cassava-leaf-disease-classification/discussion/216572",
  "author_name": "",
  "post_date": "2021-02-03T09:23:30.834707Z",
  "votes": 3,
  "comment_count": 2,
  "views": 0,
  "content": "<p>How do you collect the best list of augmentations? I personally test each agumentation and look at the local score.</p>",
  "messages": [
    {
      "id": "1183879",
      "postDate": "02/03/2021 09:23:30",
      "content": "<p>How do you collect the best list of augmentations? I personally test each agumentation and look at the local score.</p>",
      "rawMarkdown": "How do you collect the best list of augmentations? I personally test each agumentation and look at the local score.",
      "votes": null
    },
    {
      "id": "1184875",
      "postDate": "02/03/2021 20:22:07",
      "content": "<p>I use the same technique unless I know good working augmentations e.g. RandomRotation and Flipping mostly work great, especially together</p>",
      "rawMarkdown": "I use the same technique unless I know good working augmentations e.g. RandomRotation and Flipping mostly work great, especially together",
      "votes": null
    },
    {
      "id": "1186686",
      "postDate": "02/05/2021 02:01:21",
      "content": "<p>i have wondered the exact same thing.<br>\ni forked a notebook that already had a whole chain of augmentations - and despite my best effort to improve on the ones chosen (and the parameters for each), all my testing scored lower.<br>\ni did notice an interesting toolset from Alex Buslaev called \"AutoAlbument\" that actually learns ideal augmentations and tuning parameters.<br>\n<a href=\"https://albumentations.ai/docs/autoalbument/\" target=\"_blank\">https://albumentations.ai/docs/autoalbument/</a><br>\ni haven't tried it, but i am tempted, given how powerful augmentations seem to be generally.</p>",
      "rawMarkdown": "i have wondered the exact same thing.\ni forked a notebook that already had a whole chain of augmentations - and despite my best effort to improve on the ones chosen (and the parameters for each), all my testing scored lower.\ni did notice an interesting toolset from Alex Buslaev called \"AutoAlbument\" that actually learns ideal augmentations and tuning parameters.\nhttps://albumentations.ai/docs/autoalbument/\ni haven't tried it, but i am tempted, given how powerful augmentations seem to be generally.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1184875,
      "author_name": "aliabdin1",
      "author_url": "",
      "post_date": "02/03/2021 20:22:07",
      "content": "<p>I use the same technique unless I know good working augmentations e.g. RandomRotation and Flipping mostly work great, especially together</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1186686,
      "author_name": "daveccampbell",
      "author_url": "",
      "post_date": "02/05/2021 02:01:21",
      "content": "<p>i have wondered the exact same thing.<br>\ni forked a notebook that already had a whole chain of augmentations - and despite my best effort to improve on the ones chosen (and the parameters for each), all my testing scored lower.<br>\ni did notice an interesting toolset from Alex Buslaev called \"AutoAlbument\" that actually learns ideal augmentations and tuning parameters.<br>\n<a href=\"https://albumentations.ai/docs/autoalbument/\" target=\"_blank\">https://albumentations.ai/docs/autoalbument/</a><br>\ni haven't tried it, but i am tempted, given how powerful augmentations seem to be generally.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1183879": "How do you collect the best list of augmentations? I personally test each agumentation and look at the local score.",
    "1184875": "I use the same technique unless I know good working augmentations e.g. RandomRotation and Flipping mostly work great, especially together",
    "1186686": "i have wondered the exact same thing.\ni forked a notebook that already had a whole chain of augmentations - and despite my best effort to improve on the ones chosen (and the parameters for each), all my testing scored lower.\ni did notice an interesting toolset from Alex Buslaev called \"AutoAlbument\" that actually learns ideal augmentations and tuning parameters.\nhttps://albumentations.ai/docs/autoalbument/\ni haven't tried it, but i am tempted, given how powerful augmentations seem to be generally."
  },
  "source": "meta"
}