{
  "id": 313123,
  "title": "Updated Dataset size increased 3 times !!! ",
  "url": "/competitions/ultra-mnist/discussion/313123",
  "author_name": "",
  "post_date": "2022-03-15T17:58:21.360835200Z",
  "votes": 6,
  "comment_count": 3,
  "views": 0,
  "content": "<p>The new dataset is almost 3 times the previous one and contains the same number of images as the old dataset with the same distribution of 1000 images for every class from 0-27. <br>\nThe size of the old dataset was <b></b> as compared to the updated dataset which is <b></b>.</p>\n<p><b></b></p>\n<p>A sample from both updated dataset and previous dataset  : </p>\n<h4>UPDATED Dataset -</h4>\n<p><img src=\"https://raw.githubusercontent.com/sanskar-hasija/kaggle/main/images/sample_image_new.jpeg\" alt=\"\"></p>\n<h4>OLD Dataset -</h4>\n<p><img src=\"https://raw.githubusercontent.com/sanskar-hasija/kaggle/main/images/aabwhmwmfo.jpeg\" alt=\"\"></p>",
  "messages": [
    {
      "id": "1723778",
      "postDate": "03/15/2022 17:58:21",
      "content": "<p>The new dataset is almost 3 times the previous one and contains the same number of images as the old dataset with the same distribution of 1000 images for every class from 0-27. <br>\nThe size of the old dataset was <b></b> as compared to the updated dataset which is <b></b>.</p>\n<p><b></b></p>\n<p>A sample from both updated dataset and previous dataset  : </p>\n<h4>UPDATED Dataset -</h4>\n<p><img src=\"https://raw.githubusercontent.com/sanskar-hasija/kaggle/main/images/sample_image_new.jpeg\" alt=\"\"></p>\n<h4>OLD Dataset -</h4>\n<p><img src=\"https://raw.githubusercontent.com/sanskar-hasija/kaggle/main/images/aabwhmwmfo.jpeg\" alt=\"\"></p>",
      "rawMarkdown": "The new dataset is almost 3 times the previous one and contains the same number of images as the old dataset with the same distribution of 1000 images for every class from 0-27. \nThe size of the old dataset was <b><u>12.42GB</u></b> as compared to the updated dataset which is <b><u>34.5GB</u></b>.\n\n <b><u>There is a visible increase in complexity in images.</u></b>\n\nA sample from both updated dataset and previous dataset  : \n#### UPDATED Dataset - \n![](https://raw.githubusercontent.com/sanskar-hasija/kaggle/main/images/sample_image_new.jpeg)\n#### OLD Dataset - \n![](https://raw.githubusercontent.com/sanskar-hasija/kaggle/main/images/aabwhmwmfo.jpeg)",
      "votes": null
    },
    {
      "id": "1723809",
      "postDate": "03/15/2022 18:38:14",
      "content": "<p>Now, sometimes, it difficult to recognize a digit even for a person… <br>\nIt will be a great challenge to remove this background!</p>",
      "rawMarkdown": "Now, sometimes, it difficult to recognize a digit even for a person... \nIt will be a great challenge to remove this background!",
      "votes": null
    },
    {
      "id": "1724260",
      "postDate": "03/16/2022 05:52:44",
      "content": "<p>Totally agreed <a href=\"https://www.kaggle.com/dmitrylessy\" target=\"_blank\">@dmitrylessy</a> </p>",
      "rawMarkdown": "Totally agreed @dmitrylessy",
      "votes": null
    },
    {
      "id": "1731127",
      "postDate": "03/22/2022 02:24:04",
      "content": "<p><a href=\"https://www.kaggle.com/odins0n\" target=\"_blank\">@odins0n</a> it is jpeg compression behind the hood!!!</p>\n<p>Something interesting - 19GB images (same dataset as new one) when zipper may easily become 1.9GB!</p>",
      "rawMarkdown": "odins0n it is jpeg compression behind the hood!!!\n\nSomething interesting - 19GB images (same dataset as new one) when zipper may easily become 1.9GB!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1723809,
      "author_name": "dmitrylessy",
      "author_url": "",
      "post_date": "03/15/2022 18:38:14",
      "content": "<p>Now, sometimes, it difficult to recognize a digit even for a person… <br>\nIt will be a great challenge to remove this background!</p>",
      "votes": null,
      "replies": [
        {
          "id": 1724260,
          "author_name": "odins0n",
          "author_url": "",
          "post_date": "03/16/2022 05:52:44",
          "content": "<p>Totally agreed <a href=\"https://www.kaggle.com/dmitrylessy\" target=\"_blank\">@dmitrylessy</a> </p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 1731127,
      "author_name": "l0new0lf",
      "author_url": "",
      "post_date": "03/22/2022 02:24:04",
      "content": "<p><a href=\"https://www.kaggle.com/odins0n\" target=\"_blank\">@odins0n</a> it is jpeg compression behind the hood!!!</p>\n<p>Something interesting - 19GB images (same dataset as new one) when zipper may easily become 1.9GB!</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1723778": "The new dataset is almost 3 times the previous one and contains the same number of images as the old dataset with the same distribution of 1000 images for every class from 0-27. \nThe size of the old dataset was <b><u>12.42GB</u></b> as compared to the updated dataset which is <b><u>34.5GB</u></b>.\n\n <b><u>There is a visible increase in complexity in images.</u></b>\n\nA sample from both updated dataset and previous dataset  : \n#### UPDATED Dataset - \n![](https://raw.githubusercontent.com/sanskar-hasija/kaggle/main/images/sample_image_new.jpeg)\n#### OLD Dataset - \n![](https://raw.githubusercontent.com/sanskar-hasija/kaggle/main/images/aabwhmwmfo.jpeg)",
    "1723809": "Now, sometimes, it difficult to recognize a digit even for a person... \nIt will be a great challenge to remove this background!",
    "1724260": "Totally agreed @dmitrylessy",
    "1731127": "odins0n it is jpeg compression behind the hood!!!\n\nSomething interesting - 19GB images (same dataset as new one) when zipper may easily become 1.9GB!"
  },
  "source": "meta"
}