{
  "id": 170138,
  "title": "MobileNet On Melanoma",
  "url": "/competitions/siim-isic-melanoma-classification/discussion/170138",
  "author_name": "Aditya Baurai",
  "post_date": "2020-07-26T15:14:41.330000",
  "votes": 1,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Tried it with MobileNetv2 architecture using Tensorflow.</p>\n\n<p>Those who are looking to begin model training on TPU and designing efficient data pipelines on Tensorflow, do give it a shot. I have tried to explain all the components used in great detail.</p>\n\n<p>This also some cool preprocessing such as histogram equalization and denoising. I still haven't incorporated them during the model training as it was difficult to meddle with existing tfrecords. However, that will be done in time to come.</p>\n\n<p>Hope you folks enjoy it as much as I did creating this content.</p>\n\n<p><em>Do upvote if it turns out to be useful!!</em></p>\n\n<p><strong><a href=\"https://www.kaggle.com/fireheart7/melanoma-mobilenetv2?scriptVersionId=39562798\">https://www.kaggle.com/fireheart7/melanoma-mobilenetv2?scriptVersionId=39562798</a></strong></p>",
  "messages": [
    {
      "id": 946431,
      "postDate": "2020-07-26T15:14:41.330Z",
      "content": "<p>Tried it with MobileNetv2 architecture using Tensorflow.</p>\n\n<p>Those who are looking to begin model training on TPU and designing efficient data pipelines on Tensorflow, do give it a shot. I have tried to explain all the components used in great detail.</p>\n\n<p>This also some cool preprocessing such as histogram equalization and denoising. I still haven't incorporated them during the model training as it was difficult to meddle with existing tfrecords. However, that will be done in time to come.</p>\n\n<p>Hope you folks enjoy it as much as I did creating this content.</p>\n\n<p><em>Do upvote if it turns out to be useful!!</em></p>\n\n<p><strong><a href=\"https://www.kaggle.com/fireheart7/melanoma-mobilenetv2?scriptVersionId=39562798\">https://www.kaggle.com/fireheart7/melanoma-mobilenetv2?scriptVersionId=39562798</a></strong></p>",
      "rawMarkdown": "Tried it with MobileNetv2 architecture using Tensorflow.\n\nThose who are looking to begin model training on TPU and designing efficient data pipelines on Tensorflow, do give it a shot. I have tried to explain all the components used in great detail.\n\nThis also some cool preprocessing such as histogram equalization and denoising. I still haven't incorporated them during the model training as it was difficult to meddle with existing tfrecords. However, that will be done in time to come.\n\nHope you folks enjoy it as much as I did creating this content.\n\n*Do upvote if it turns out to be useful!!*\n\n**https://www.kaggle.com/fireheart7/melanoma-mobilenetv2?scriptVersionId=39562798**\n\n",
      "votes": 2
    },
    {
      "id": 947183,
      "postDate": "2020-07-27T05:51:17.890Z",
      "content": "<p>Nice work! I used MobileNet on the past due to efficiency. Will check out this kernel</p>",
      "rawMarkdown": "Nice work! I used MobileNet on the past due to efficiency. Will check out this kernel"
    },
    {
      "id": 946666,
      "postDate": "2020-07-26T18:03:14.643Z",
      "rawMarkdown": "",
      "isDeleted": true,
      "replies": [
        {
          "id": 947330,
          "postDate": "2020-07-27T07:30:32.537Z",
          "content": "<p>Yes!! Even I was stunned!</p>",
          "rawMarkdown": "Yes!! Even I was stunned!"
        }
      ]
    }
  ],
  "comments": [
    {
      "id": 947183,
      "author_name": "André Neves",
      "author_url": "",
      "post_date": "2020-07-27T05:51:17.890000",
      "content": "<p>Nice work! I used MobileNet on the past due to efficiency. Will check out this kernel</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 946666,
      "author_name": "",
      "author_url": "",
      "post_date": "2020-07-26T18:03:14.643000",
      "content": "",
      "votes": 0,
      "replies": [
        {
          "id": 947330,
          "author_name": "Aditya Baurai",
          "author_url": "",
          "post_date": "2020-07-27T07:30:32.537000",
          "content": "<p>Yes!! Even I was stunned!</p>",
          "votes": 0,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "946431": "Tried it with MobileNetv2 architecture using Tensorflow.\n\nThose who are looking to begin model training on TPU and designing efficient data pipelines on Tensorflow, do give it a shot. I have tried to explain all the components used in great detail.\n\nThis also some cool preprocessing such as histogram equalization and denoising. I still haven't incorporated them during the model training as it was difficult to meddle with existing tfrecords. However, that will be done in time to come.\n\nHope you folks enjoy it as much as I did creating this content.\n\n*Do upvote if it turns out to be useful!!*\n\n**https://www.kaggle.com/fireheart7/melanoma-mobilenetv2?scriptVersionId=39562798**\n\n",
    "947183": "Nice work! I used MobileNet on the past due to efficiency. Will check out this kernel",
    "946666": ""
  }
}