{
  "id": 31472,
  "title": "Extracted coordinates available here :-)",
  "url": "/competitions/noaa-fisheries-steller-sea-lion-population-count/discussion/31472",
  "author_name": "",
  "post_date": "2017-04-11T22:09:40.709679800Z",
  "votes": 26,
  "comment_count": 8,
  "views": 0,
  "content": "<p>Hi,</p>\n\n<p>I wrote a script based on the code from Radu and ran over the full training set to obtain dot coordinates and I'm sharing the results. Sooo, without further ado:</p>\n\n<p><a href=\"https://raw.githubusercontent.com/cxz/noaa/master/data/coords.csv\">https://raw.githubusercontent.com/cxz/noaa/master/data/coords.csv</a></p>\n\n<p>There are a about ~40 missing images for several reasons, so expect this file to be updated later.  Image Train/912.jpg contains red drawings, 530 and 630 have wrong dimensions (they're flipped), and a few others need a different method for extracting the blobs as they cause the scikit-image library to hang indefinitely (basically the routine that does blob prunning is quadratic and the images contains a lot of 1-pixel differences due to compression artifacts, etc).</p>\n\n<p>As people have noted, the counts will not match for every image, it's up to each one to decide how to handle that.</p>\n\n<p>This is not oficial and not final. If anyone wants to help let's make it collaborative and update the repository with fixes.</p>\n\n<p>Enjoy!</p>",
  "messages": [
    {
      "id": "174537",
      "postDate": "04/11/2017 22:09:40",
      "content": "<p>Hi,</p>\n\n<p>I wrote a script based on the code from Radu and ran over the full training set to obtain dot coordinates and I'm sharing the results. Sooo, without further ado:</p>\n\n<p><a href=\"https://raw.githubusercontent.com/cxz/noaa/master/data/coords.csv\">https://raw.githubusercontent.com/cxz/noaa/master/data/coords.csv</a></p>\n\n<p>There are a about ~40 missing images for several reasons, so expect this file to be updated later.  Image Train/912.jpg contains red drawings, 530 and 630 have wrong dimensions (they're flipped), and a few others need a different method for extracting the blobs as they cause the scikit-image library to hang indefinitely (basically the routine that does blob prunning is quadratic and the images contains a lot of 1-pixel differences due to compression artifacts, etc).</p>\n\n<p>As people have noted, the counts will not match for every image, it's up to each one to decide how to handle that.</p>\n\n<p>This is not oficial and not final. If anyone wants to help let's make it collaborative and update the repository with fixes.</p>\n\n<p>Enjoy!</p>",
      "rawMarkdown": "Hi,\n\nI wrote a script based on the code from Radu and ran over the full training set to obtain dot coordinates and I'm sharing the results. Sooo, without further ado:\n\n[https://raw.githubusercontent.com/cxz/noaa/master/data/coords.csv][1]\n\nThere are a about ~40 missing images for several reasons, so expect this file to be updated later.  Image Train/912.jpg contains red drawings, 530 and 630 have wrong dimensions (they're flipped), and a few others need a different method for extracting the blobs as they cause the scikit-image library to hang indefinitely (basically the routine that does blob prunning is quadratic and the images contains a lot of 1-pixel differences due to compression artifacts, etc).\n\nAs people have noted, the counts will not match for every image, it's up to each one to decide how to handle that.\n\nThis is not oficial and not final. If anyone wants to help let's make it collaborative and update the repository with fixes.\n\nEnjoy!\n\n  [1]: https://raw.githubusercontent.com/cxz/noaa/master/data/coords.csv",
      "votes": null
    },
    {
      "id": "175370",
      "postDate": "04/15/2017 05:12:27",
      "content": "<p>For some images I get over 1000 coordinates while the number of sea lions is well below 200. Am I missing something ?</p>",
      "rawMarkdown": "For some images I get over 1000 coordinates while the number of sea lions is well below 200. Am I missing something ?",
      "votes": null
    },
    {
      "id": "175541",
      "postDate": "04/16/2017 02:52:14",
      "content": "<p>Yes, image 946 for example is completely wrong. I'm filtering out images for which the script didn't work correctly for now. For example: 34, 491, 839, 840, 882, 912 and 946.</p>",
      "rawMarkdown": "Yes, image 946 for example is completely wrong. I'm filtering out images for which the script didn't work correctly for now. For example: 34, 491, 839, 840, 882, 912 and 946.",
      "votes": null
    },
    {
      "id": "175720",
      "postDate": "04/17/2017 06:33:16",
      "content": "<p>This helps a lot man!</p>",
      "rawMarkdown": "This helps a lot man!",
      "votes": null
    },
    {
      "id": "175956",
      "postDate": "04/18/2017 08:48:55",
      "content": "<p>Nice work! very helpful to me!!</p>",
      "rawMarkdown": "Nice work! very helpful to me!!",
      "votes": null
    },
    {
      "id": "179042",
      "postDate": "04/29/2017 21:38:50",
      "content": "<p>Any chance you can post the source to your script?  Thanks!</p>",
      "rawMarkdown": "Any chance you can post the source to your script?  Thanks!",
      "votes": null
    },
    {
      "id": "179063",
      "postDate": "04/30/2017 01:32:47",
      "content": "<p>Hi,\nThere's an equivalent in the kernels:\n <a href=\"https://www.kaggle.com/threeplusone/noaa-fisheries-steller-sea-lion-population-count/sea-lion-coordinates\">https://www.kaggle.com/threeplusone/noaa-fisheries-steller-sea-lion-population-count/sea-lion-coordinates</a></p>",
      "rawMarkdown": "Hi,\nThere's an equivalent in the kernels:\n https://www.kaggle.com/threeplusone/noaa-fisheries-steller-sea-lion-population-count/sea-lion-coordinates",
      "votes": null
    },
    {
      "id": "179075",
      "postDate": "04/30/2017 03:10:16",
      "content": "<p>very nice!</p>",
      "rawMarkdown": "very nice!",
      "votes": null
    },
    {
      "id": "180529",
      "postDate": "05/05/2017 19:22:33",
      "content": "<p>Thanks amaia!    In your description, I believe it is 638 which is flipped/transposed, not 630.   </p>",
      "rawMarkdown": "Thanks amaia!    In your description, I believe it is 638 which is flipped/transposed, not 630.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 175370,
      "author_name": "zaidalyafeai",
      "author_url": "",
      "post_date": "04/15/2017 05:12:27",
      "content": "<p>For some images I get over 1000 coordinates while the number of sea lions is well below 200. Am I missing something ?</p>",
      "votes": null,
      "replies": [
        {
          "id": 175541,
          "author_name": "aamaia",
          "author_url": "",
          "post_date": "04/16/2017 02:52:14",
          "content": "<p>Yes, image 946 for example is completely wrong. I'm filtering out images for which the script didn't work correctly for now. For example: 34, 491, 839, 840, 882, 912 and 946.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 175720,
      "author_name": "kingburrito666",
      "author_url": "",
      "post_date": "04/17/2017 06:33:16",
      "content": "<p>This helps a lot man!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 175956,
      "author_name": "kuo77122",
      "author_url": "",
      "post_date": "04/18/2017 08:48:55",
      "content": "<p>Nice work! very helpful to me!!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 179042,
      "author_name": "happycube",
      "author_url": "",
      "post_date": "04/29/2017 21:38:50",
      "content": "<p>Any chance you can post the source to your script?  Thanks!</p>",
      "votes": null,
      "replies": [
        {
          "id": 179063,
          "author_name": "aamaia",
          "author_url": "",
          "post_date": "04/30/2017 01:32:47",
          "content": "<p>Hi,\nThere's an equivalent in the kernels:\n <a href=\"https://www.kaggle.com/threeplusone/noaa-fisheries-steller-sea-lion-population-count/sea-lion-coordinates\">https://www.kaggle.com/threeplusone/noaa-fisheries-steller-sea-lion-population-count/sea-lion-coordinates</a></p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 179075,
      "author_name": "s19g92",
      "author_url": "",
      "post_date": "04/30/2017 03:10:16",
      "content": "<p>very nice!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 180529,
      "author_name": "sasrdw",
      "author_url": "",
      "post_date": "05/05/2017 19:22:33",
      "content": "<p>Thanks amaia!    In your description, I believe it is 638 which is flipped/transposed, not 630.   </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "174537": "Hi,\n\nI wrote a script based on the code from Radu and ran over the full training set to obtain dot coordinates and I'm sharing the results. Sooo, without further ado:\n\n[https://raw.githubusercontent.com/cxz/noaa/master/data/coords.csv][1]\n\nThere are a about ~40 missing images for several reasons, so expect this file to be updated later.  Image Train/912.jpg contains red drawings, 530 and 630 have wrong dimensions (they're flipped), and a few others need a different method for extracting the blobs as they cause the scikit-image library to hang indefinitely (basically the routine that does blob prunning is quadratic and the images contains a lot of 1-pixel differences due to compression artifacts, etc).\n\nAs people have noted, the counts will not match for every image, it's up to each one to decide how to handle that.\n\nThis is not oficial and not final. If anyone wants to help let's make it collaborative and update the repository with fixes.\n\nEnjoy!\n\n  [1]: https://raw.githubusercontent.com/cxz/noaa/master/data/coords.csv",
    "175370": "For some images I get over 1000 coordinates while the number of sea lions is well below 200. Am I missing something ?",
    "175541": "Yes, image 946 for example is completely wrong. I'm filtering out images for which the script didn't work correctly for now. For example: 34, 491, 839, 840, 882, 912 and 946.",
    "175720": "This helps a lot man!",
    "175956": "Nice work! very helpful to me!!",
    "179042": "Any chance you can post the source to your script?  Thanks!",
    "179063": "Hi,\nThere's an equivalent in the kernels:\n https://www.kaggle.com/threeplusone/noaa-fisheries-steller-sea-lion-population-count/sea-lion-coordinates",
    "179075": "very nice!",
    "180529": "Thanks amaia!    In your description, I believe it is 638 which is flipped/transposed, not 630."
  },
  "source": "meta"
}