{
  "id": 171968,
  "title": "Dataset(external) simplified to 256,384,512,784 for ensemble",
  "url": "/competitions/siim-isic-melanoma-classification/discussion/171968",
  "author_name": "",
  "post_date": "2020-08-03T07:40:50.837901700Z",
  "votes": 2,
  "comment_count": 1,
  "views": 0,
  "content": "<p>As I have seen from the competitions discussion forums, The ensemble models trained on different image sizes are working pretty nice. So I planned to ease workflows of mine on arranging datasets to 256,384,512,784 and share them with the group. The dataset prepared below are taken from <a href=\"/cdeotte\">@cdeotte</a>'s tfrecords dataset and it was processed by <a href=\"/tahsin\">@tahsin</a> . You will find the <a href=\"/tahsin\">@tahsin</a> 's dataset <a href=\"https://www.kaggle.com/tahsin/tfrecords-converted-to-jpeg-and-meta-data\">here</a>. </p>\n\n<p>The datasets uploaded by me:\nMelanoma external 256 jpg :-<a href=\"https://www.kaggle.com/tawheedrony/melanoma-jpg-external-256\">Melanoma external 256 jpg</a>\nMelanoma external 384 jpg :-<a href=\"https://www.kaggle.com/tawheedrony/melanoma-jpg-external\">Melanoma external 384 jpg</a>\nMelanoma external 384 jpg :- Will upload later\nMelanoma external 784 jpg :- Use <a href=\"https://www.kaggle.com/tahsin/tfrecords-converted-to-jpeg-and-meta-data\">this dataset</a>. </p>\n\n<p>I am a newbie just starting to explore things in this domain. Help me grow by pointing out my mistakes. Thanks everyone.</p>",
  "messages": [
    {
      "id": "956053",
      "postDate": "08/03/2020 07:40:50",
      "content": "<p>As I have seen from the competitions discussion forums, The ensemble models trained on different image sizes are working pretty nice. So I planned to ease workflows of mine on arranging datasets to 256,384,512,784 and share them with the group. The dataset prepared below are taken from <a href=\"/cdeotte\">@cdeotte</a>'s tfrecords dataset and it was processed by <a href=\"/tahsin\">@tahsin</a> . You will find the <a href=\"/tahsin\">@tahsin</a> 's dataset <a href=\"https://www.kaggle.com/tahsin/tfrecords-converted-to-jpeg-and-meta-data\">here</a>. </p>\n\n<p>The datasets uploaded by me:\nMelanoma external 256 jpg :-<a href=\"https://www.kaggle.com/tawheedrony/melanoma-jpg-external-256\">Melanoma external 256 jpg</a>\nMelanoma external 384 jpg :-<a href=\"https://www.kaggle.com/tawheedrony/melanoma-jpg-external\">Melanoma external 384 jpg</a>\nMelanoma external 384 jpg :- Will upload later\nMelanoma external 784 jpg :- Use <a href=\"https://www.kaggle.com/tahsin/tfrecords-converted-to-jpeg-and-meta-data\">this dataset</a>. </p>\n\n<p>I am a newbie just starting to explore things in this domain. Help me grow by pointing out my mistakes. Thanks everyone.</p>",
      "rawMarkdown": "As I have seen from the competitions discussion forums, The ensemble models trained on different image sizes are working pretty nice. So I planned to ease workflows of mine on arranging datasets to 256,384,512,784 and share them with the group. The dataset prepared below are taken from @cdeotte's tfrecords dataset and it was processed by @tahsin . You will find the @tahsin 's dataset [here](https://www.kaggle.com/tahsin/tfrecords-converted-to-jpeg-and-meta-data). \n\nThe datasets uploaded by me:\nMelanoma external 256 jpg :-[Melanoma external 256 jpg](https://www.kaggle.com/tawheedrony/melanoma-jpg-external-256)\nMelanoma external 384 jpg :-[Melanoma external 384 jpg](https://www.kaggle.com/tawheedrony/melanoma-jpg-external)\nMelanoma external 384 jpg :- Will upload later\nMelanoma external 784 jpg :- Use [this dataset](https://www.kaggle.com/tahsin/tfrecords-converted-to-jpeg-and-meta-data). \n\nI am a newbie just starting to explore things in this domain. Help me grow by pointing out my mistakes. Thanks everyone.",
      "votes": null
    },
    {
      "id": "956057",
      "postDate": "08/03/2020 07:45:18",
      "content": "<p>Thanks for sharing your work.</p>",
      "rawMarkdown": "Thanks for sharing your work.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 956057,
      "author_name": "fchmiel",
      "author_url": "",
      "post_date": "08/03/2020 07:45:18",
      "content": "<p>Thanks for sharing your work.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "956053": "As I have seen from the competitions discussion forums, The ensemble models trained on different image sizes are working pretty nice. So I planned to ease workflows of mine on arranging datasets to 256,384,512,784 and share them with the group. The dataset prepared below are taken from @cdeotte's tfrecords dataset and it was processed by @tahsin . You will find the @tahsin 's dataset [here](https://www.kaggle.com/tahsin/tfrecords-converted-to-jpeg-and-meta-data). \n\nThe datasets uploaded by me:\nMelanoma external 256 jpg :-[Melanoma external 256 jpg](https://www.kaggle.com/tawheedrony/melanoma-jpg-external-256)\nMelanoma external 384 jpg :-[Melanoma external 384 jpg](https://www.kaggle.com/tawheedrony/melanoma-jpg-external)\nMelanoma external 384 jpg :- Will upload later\nMelanoma external 784 jpg :- Use [this dataset](https://www.kaggle.com/tahsin/tfrecords-converted-to-jpeg-and-meta-data). \n\nI am a newbie just starting to explore things in this domain. Help me grow by pointing out my mistakes. Thanks everyone.",
    "956057": "Thanks for sharing your work."
  },
  "source": "meta"
}