{
  "id": 12960,
  "title": "Crop circles",
  "url": "/competitions/diabetic-retinopathy-detection/discussion/12960",
  "author_name": "",
  "post_date": "2015-03-21T16:30:20.270Z",
  "votes": null,
  "comment_count": 3,
  "views": 3082,
  "content": "<p>What's a quick way to determine the circular boundary of the actual image? &nbsp;It would make it much easier to normalize the image sizes to have that. &nbsp;Commands like &nbsp;&quot;convert -trim -fuzz '1%' test/10001_left.jpeg /tmp/tst.jpg&quot; kind of work, but with some images the fuzz factor needs to be much higher, whereas with darker images a high fuzz factor causes cropping of the actual image circle.</p>",
  "messages": [
    {
      "id": "67535",
      "postDate": "03/21/2015 16:30:20",
      "content": "<p>What's a quick way to determine the circular boundary of the actual image? &nbsp;It would make it much easier to normalize the image sizes to have that. &nbsp;Commands like &nbsp;&quot;convert -trim -fuzz '1%' test/10001_left.jpeg /tmp/tst.jpg&quot; kind of work, but with some images the fuzz factor needs to be much higher, whereas with darker images a high fuzz factor causes cropping of the actual image circle.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "69354",
      "postDate": "04/01/2015 19:17:55",
      "content": "<p>throwaway, I'm wondering if you've teamed up with anybody in this competition yet. I'm new to Machine Learning, and like you, this is my first Kaggle competition.&nbsp;I come from an original background in Physics with a relatively extensive range of work experience in computer graphics, image processing, and computer vision. At this point, I'm still working on my own, but I'd like to develop some friendship and collegiality with a few others so that we might band together as things progress.</p>\n<p>Your postings caught my eye, so if this prospect sounds interesting to you, please let me know.</p>\n<p>-moonman</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "71503",
      "postDate": "04/14/2015 14:18:55",
      "content": "<p>Hi throwaway,</p>\n<p>1) You can use ImageMagick to crop the images as in this post:&nbsp;http://www.kaggle.com/c/diabetic-retinopathy-detection/forums/t/12842/starter-code-for-leaderboard-score-of-0-46</p>\n<p>2) You can use OpenCV to crop the images as in the following code (see getHistogramEqualizeCropped):</p>\n<p>https://github.com/adarshvjois/image_feature_extractor/blob/master/ImageFeatureExtractor/src/FeatureVec.cpp</p>\n<p>Hope this helps.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "77636",
      "postDate": "05/08/2015 16:58:43",
      "content": "<p>Thanks for the suggestions Kshitiz. &nbsp;</p>\n\n<p>moonman, I am still getting going, not ready to team up yet.</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 69354,
      "author_name": "moonman1",
      "author_url": "",
      "post_date": "04/01/2015 19:17:55",
      "content": "<p>throwaway, I'm wondering if you've teamed up with anybody in this competition yet. I'm new to Machine Learning, and like you, this is my first Kaggle competition.&nbsp;I come from an original background in Physics with a relatively extensive range of work experience in computer graphics, image processing, and computer vision. At this point, I'm still working on my own, but I'd like to develop some friendship and collegiality with a few others so that we might band together as things progress.</p>\n<p>Your postings caught my eye, so if this prospect sounds interesting to you, please let me know.</p>\n<p>-moonman</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 71503,
      "author_name": "kshitizsethia",
      "author_url": "",
      "post_date": "04/14/2015 14:18:55",
      "content": "<p>Hi throwaway,</p>\n<p>1) You can use ImageMagick to crop the images as in this post:&nbsp;http://www.kaggle.com/c/diabetic-retinopathy-detection/forums/t/12842/starter-code-for-leaderboard-score-of-0-46</p>\n<p>2) You can use OpenCV to crop the images as in the following code (see getHistogramEqualizeCropped):</p>\n<p>https://github.com/adarshvjois/image_feature_extractor/blob/master/ImageFeatureExtractor/src/FeatureVec.cpp</p>\n<p>Hope this helps.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 77636,
      "author_name": "alexcoventry",
      "author_url": "",
      "post_date": "05/08/2015 16:58:43",
      "content": "<p>Thanks for the suggestions Kshitiz. &nbsp;</p>\n\n<p>moonman, I am still getting going, not ready to team up yet.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "67535": "",
    "69354": "",
    "71503": "",
    "77636": ""
  },
  "source": "meta"
}