{
  "id": 312091,
  "title": "Black AND rescaled images",
  "url": "/competitions/ultra-mnist/discussion/312091",
  "author_name": "",
  "post_date": "2022-03-10T10:40:25.100231600Z",
  "votes": 5,
  "comment_count": 1,
  "views": 0,
  "content": "<p>If anyone is interested in experimenting on resized data that has cleaned background (h/t <a href=\"https://www.kaggle.com/remekkinas\" target=\"_blank\">@remekkinas</a> for his dataset - brilliant in its simplicity of the underlying idea), you can find the kernel here:</p>\n<ul>\n<li>takes the black &amp; white data as input instead of the original one\\</li>\n<li>creates a rescaled version </li>\n<li>parallelized with joblib =&gt; faster ;-)</li>\n</ul>\n<p><a href=\"https://www.kaggle.com/konradb/fast-image-scaling-with-joblib-black\" target=\"_blank\">https://www.kaggle.com/konradb/fast-image-scaling-with-joblib-black</a></p>\n<p>Have fun.</p>",
  "messages": [
    {
      "id": "1717919",
      "postDate": "03/10/2022 10:40:25",
      "content": "<p>If anyone is interested in experimenting on resized data that has cleaned background (h/t <a href=\"https://www.kaggle.com/remekkinas\" target=\"_blank\">@remekkinas</a> for his dataset - brilliant in its simplicity of the underlying idea), you can find the kernel here:</p>\n<ul>\n<li>takes the black &amp; white data as input instead of the original one\\</li>\n<li>creates a rescaled version </li>\n<li>parallelized with joblib =&gt; faster ;-)</li>\n</ul>\n<p><a href=\"https://www.kaggle.com/konradb/fast-image-scaling-with-joblib-black\" target=\"_blank\">https://www.kaggle.com/konradb/fast-image-scaling-with-joblib-black</a></p>\n<p>Have fun.</p>",
      "rawMarkdown": "If anyone is interested in experimenting on resized data that has cleaned background (h/t @remekkinas for his dataset - brilliant in its simplicity of the underlying idea), you can find the kernel here:\n- takes the black & white data as input instead of the original one\\\n- creates a rescaled version \n- parallelized with joblib => faster ;-)\n\nhttps://www.kaggle.com/konradb/fast-image-scaling-with-joblib-black\n\nHave fun.",
      "votes": null
    },
    {
      "id": "1718116",
      "postDate": "03/10/2022 14:17:07",
      "content": "<p>Thank you <a href=\"https://www.kaggle.com/konradb\" target=\"_blank\">@konradb</a> for mentioning my work. 🙏🙏🙏<br>\nThank you for sharing joblib tip. I will update my notebook for Dataset generation. Currently I do not use it … so had to wait hours to generate all data … Will let you know about gain I get. </p>",
      "rawMarkdown": "Thank you @konradb for mentioning my work. 🙏🙏🙏\nThank you for sharing joblib tip. I will update my notebook for Dataset generation. Currently I do not use it ... so had to wait hours to generate all data ... Will let you know about gain I get.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1718116,
      "author_name": "remekkinas",
      "author_url": "",
      "post_date": "03/10/2022 14:17:07",
      "content": "<p>Thank you <a href=\"https://www.kaggle.com/konradb\" target=\"_blank\">@konradb</a> for mentioning my work. 🙏🙏🙏<br>\nThank you for sharing joblib tip. I will update my notebook for Dataset generation. Currently I do not use it … so had to wait hours to generate all data … Will let you know about gain I get. </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1717919": "If anyone is interested in experimenting on resized data that has cleaned background (h/t @remekkinas for his dataset - brilliant in its simplicity of the underlying idea), you can find the kernel here:\n- takes the black & white data as input instead of the original one\\\n- creates a rescaled version \n- parallelized with joblib => faster ;-)\n\nhttps://www.kaggle.com/konradb/fast-image-scaling-with-joblib-black\n\nHave fun.",
    "1718116": "Thank you @konradb for mentioning my work. 🙏🙏🙏\nThank you for sharing joblib tip. I will update my notebook for Dataset generation. Currently I do not use it ... so had to wait hours to generate all data ... Will let you know about gain I get."
  },
  "source": "meta"
}