{
  "id": 53410,
  "title": "How to use keras image augmentation in matterport's maskrcnn?",
  "url": "/competitions/data-science-bowl-2018/discussion/53410",
  "author_name": "Caibimiao",
  "post_date": "2018-03-30T07:35:57.274000",
  "votes": 0,
  "comment_count": 6,
  "views": 0,
  "content": "<p>The function <code>ImageDataGenerator</code> is easy to use for image augmentation in keras </p>\n\n<p>But after check the code of data loaders in matterport's maskrcnn , I found that I can't use <code>ImageDataGenerator</code>  directly, I need to implement them in function <code>load_image_gt()</code>(model.py)</p>\n\n<p>Any one has a better idea?  :-)</p>",
  "messages": [
    {
      "id": 306549,
      "postDate": "2018-03-30T16:06:43.147Z",
      "content": "<p>I am not sure about keras image augmentation. But you can find a excellent augmentation tool that can be used with any python framework in this repo: <a href=\"https://github.com/aleju/imgaug\">https://github.com/aleju/imgaug</a></p>",
      "rawMarkdown": "I am not sure about keras image augmentation. But you can find a excellent augmentation tool that can be used with any python framework in this repo: https://github.com/aleju/imgaug",
      "votes": 1,
      "replies": [
        {
          "id": 306905,
          "postDate": "2018-03-31T09:32:01.587Z",
          "content": "<p>sorry delete me</p>",
          "rawMarkdown": "sorry delete me"
        },
        {
          "id": 306907,
          "postDate": "2018-03-31T09:32:54.687Z",
          "content": "<p>yes！This is what I do now ,I use this framework and put them into the <code>load_image_gt()</code> function\nBut I think there should be a better way here with keras\nThank you！</p>",
          "rawMarkdown": "yes！This is what I do now ,I use this framework and put them into the `load_image_gt()` function\nBut I think there should be a better way here with keras\nThank you！"
        }
      ]
    },
    {
      "id": 307011,
      "postDate": "2018-03-31T15:45:35.760Z",
      "content": "<p>I post the information <a href=\"https://www.kaggle.com/c/data-science-bowl-2018/discussion/53495\">here</a></p>\n\n<p>You can try it.</p>",
      "rawMarkdown": "I post the information [here][1]\n\nYou can try it.\n\n  [1]: https://www.kaggle.com/c/data-science-bowl-2018/discussion/53495",
      "replies": [
        {
          "id": 307203,
          "postDate": "2018-04-01T04:34:58.663Z",
          "content": "<p>Thx a lot!</p>",
          "rawMarkdown": "Thx a lot!"
        }
      ]
    },
    {
      "id": 306294,
      "postDate": "2018-03-30T07:35:57.273Z",
      "content": "<p>The function <code>ImageDataGenerator</code> is easy to use for image augmentation in keras </p>\n\n<p>But after check the code of data loaders in matterport's maskrcnn , I found that I can't use <code>ImageDataGenerator</code>  directly, I need to implement them in function <code>load_image_gt()</code>(model.py)</p>\n\n<p>Any one has a better idea?  :-)</p>",
      "rawMarkdown": "The function `ImageDataGenerator` is easy to use for image augmentation in keras \n\nBut after check the code of data loaders in matterport's maskrcnn , I found that I can't use `ImageDataGenerator`  directly, I need to implement them in function `load_image_gt()`(model.py)\n\nAny one has a better idea?  :-)"
    },
    {
      "id": 678371,
      "postDate": "2019-11-21T09:38:15.473Z",
      "rawMarkdown": "",
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 306549,
      "author_name": "MariGaneshKumar",
      "author_url": "",
      "post_date": "2018-03-30T16:06:43.147000",
      "content": "<p>I am not sure about keras image augmentation. But you can find a excellent augmentation tool that can be used with any python framework in this repo: <a href=\"https://github.com/aleju/imgaug\">https://github.com/aleju/imgaug</a></p>",
      "votes": 1,
      "replies": [
        {
          "id": 306905,
          "author_name": "bintan_FDU",
          "author_url": "",
          "post_date": "2018-03-31T09:32:01.587000",
          "content": "<p>sorry delete me</p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 306907,
          "author_name": "Caibimiao",
          "author_url": "",
          "post_date": "2018-03-31T09:32:54.687000",
          "content": "<p>yes！This is what I do now ,I use this framework and put them into the <code>load_image_gt()</code> function\nBut I think there should be a better way here with keras\nThank you！</p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 307011,
      "author_name": "IAmLegend",
      "author_url": "",
      "post_date": "2018-03-31T15:45:35.760000",
      "content": "<p>I post the information <a href=\"https://www.kaggle.com/c/data-science-bowl-2018/discussion/53495\">here</a></p>\n\n<p>You can try it.</p>",
      "votes": 0,
      "replies": [
        {
          "id": 307203,
          "author_name": "Caibimiao",
          "author_url": "",
          "post_date": "2018-04-01T04:34:58.663000",
          "content": "<p>Thx a lot!</p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 678371,
      "author_name": "",
      "author_url": "",
      "post_date": "2019-11-21T09:38:15.473000",
      "content": "",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "306549": "I am not sure about keras image augmentation. But you can find a excellent augmentation tool that can be used with any python framework in this repo: https://github.com/aleju/imgaug",
    "307011": "I post the information [here][1]\n\nYou can try it.\n\n  [1]: https://www.kaggle.com/c/data-science-bowl-2018/discussion/53495",
    "306294": "The function `ImageDataGenerator` is easy to use for image augmentation in keras \n\nBut after check the code of data loaders in matterport's maskrcnn , I found that I can't use `ImageDataGenerator`  directly, I need to implement them in function `load_image_gt()`(model.py)\n\nAny one has a better idea?  :-)",
    "678371": ""
  }
}