{
  "id": 30742,
  "title": "Welcome!",
  "url": "/competitions/noaa-fisheries-steller-sea-lion-population-count/discussion/30742",
  "author_name": "",
  "post_date": "2017-03-27T21:13:56.993683400Z",
  "votes": 4,
  "comment_count": 15,
  "views": 0,
  "content": "<p>Welcome everyone to our latest competition! I hope you enjoy it.</p>\n\n<p>Because this is a large dataset, we'll be making it available via BitTorrent. I'm setting that up now.</p>",
  "messages": [
    {
      "id": "170867",
      "postDate": "03/27/2017 21:13:56",
      "content": "<p>Welcome everyone to our latest competition! I hope you enjoy it.</p>\n\n<p>Because this is a large dataset, we'll be making it available via BitTorrent. I'm setting that up now.</p>",
      "rawMarkdown": "Welcome everyone to our latest competition! I hope you enjoy it.\n\nBecause this is a large dataset, we'll be making it available via BitTorrent. I'm setting that up now.",
      "votes": null
    },
    {
      "id": "170886",
      "postDate": "03/27/2017 22:17:48",
      "content": "<p>Is it possible to have just bounding boxes as annotations in text file? The dotted image is hard to use directly. BBox texts are much smaller too.</p>\n\n<p>for example, <a href=\"http://mscoco.org/dataset/#download\">http://mscoco.org/dataset/#download</a></p>\n\n<p>It really makes no sense to provide duplicated images just for annotations.</p>",
      "rawMarkdown": "Is it possible to have just bounding boxes as annotations in text file? The dotted image is hard to use directly. BBox texts are much smaller too.\n\nfor example, http://mscoco.org/dataset/#download\n\nIt really makes no sense to provide duplicated images just for annotations.",
      "votes": null
    },
    {
      "id": "170900",
      "postDate": "03/27/2017 22:52:15",
      "content": "<p>I would second this motion.  Although, I'm guessing that it would be just x,y coordinates per dot.</p>",
      "rawMarkdown": "I would second this motion.  Although, I'm guessing that it would be just x,y coordinates per dot.",
      "votes": null
    },
    {
      "id": "170901",
      "postDate": "03/27/2017 23:02:01",
      "content": "<p>Bounding boxes would be nice, but we don't have the capacity to generate them. I'll be enabling kernels on this competition shortly, though, so people can share their data ingest recipes.</p>",
      "rawMarkdown": "Bounding boxes would be nice, but we don't have the capacity to generate them. I'll be enabling kernels on this competition shortly, though, so people can share their data ingest recipes.",
      "votes": null
    },
    {
      "id": "170957",
      "postDate": "03/28/2017 05:26:11",
      "content": "",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "170958",
      "postDate": "03/28/2017 05:27:53",
      "content": "<p>I think having a sheet of </p>\n\n<p>train_id, dot_color, x_cord, y_cord</p>\n\n<p>would be greatly helpful. This would make data ingesting much easier, and cut the train set size in half by removing the need of the annotated images.</p>",
      "rawMarkdown": "I think having a sheet of \n\ntrain_id, dot_color, x_cord, y_cord\n \nwould be greatly helpful. This would make data ingesting much easier, and cut the train set size in half by removing the need of the annotated images.",
      "votes": null
    },
    {
      "id": "171000",
      "postDate": "03/28/2017 09:40:14",
      "content": "<p>I double that. We need just text file with dots location. In other case everyone will need to make their own parser of images to find these dots.</p>",
      "rawMarkdown": "I double that. We need just text file with dots location. In other case everyone will need to make their own parser of images to find these dots.",
      "votes": null
    },
    {
      "id": "171159",
      "postDate": "03/28/2017 21:55:45",
      "content": "<p>I wonder why there is no test data available in kernels or I am missing something ?</p>",
      "rawMarkdown": "I wonder why there is no test data available in kernels or I am missing something ?",
      "votes": null
    },
    {
      "id": "171368",
      "postDate": "03/29/2017 16:55:39",
      "content": "<p>I have written this notebook to get somehow reliable coordinates of the dots:\n<a href=\"https://www.kaggle.com/mohamedsamy/noaa-fisheries-steller-sea-lion-population-count/generating-text-files-for-dots-coordinates\">https://www.kaggle.com/mohamedsamy/noaa-fisheries-steller-sea-lion-population-count/generating-text-files-for-dots-coordinates</a>\nI have tested it for a couple of images, and I find it pretty good till more robust labels are generated.</p>",
      "rawMarkdown": "I have written this notebook to get somehow reliable coordinates of the dots:\nhttps://www.kaggle.com/mohamedsamy/noaa-fisheries-steller-sea-lion-population-count/generating-text-files-for-dots-coordinates\nI have tested it for a couple of images, and I find it pretty good till more robust labels are generated.",
      "votes": null
    },
    {
      "id": "171403",
      "postDate": "03/29/2017 19:49:59",
      "content": "<p>Hi, my understanding is that the Train and TrainDotted images should be the same except for the dots and the black regions in TrainDotted. I'm looking at the small train dataset of 11 images and it seems Train/7.jpg and TrainDotted/7.jpg are mirrored. There is also something weird going on with Train/9.jpg and TrainDotted/9.jpg. They show slightly different areas, so there is no close similarity between the two images. DataCanary, would it be possible to check the other train images and fix these problems?</p>",
      "rawMarkdown": "Hi, my understanding is that the Train and TrainDotted images should be the same except for the dots and the black regions in TrainDotted. I'm looking at the small train dataset of 11 images and it seems Train/7.jpg and TrainDotted/7.jpg are mirrored. There is also something weird going on with Train/9.jpg and TrainDotted/9.jpg. They show slightly different areas, so there is no close similarity between the two images. DataCanary, would it be possible to check the other train images and fix these problems?",
      "votes": null
    },
    {
      "id": "171513",
      "postDate": "03/30/2017 11:24:04",
      "content": "<p>Yes, i also found these 3 images that are not exact copies </p>\n\n<p><a href=\"https://www.kaggle.com/vfdev5/noaa-fisheries-steller-sea-lion-population-count/traindotted-train-for-3-7-9-jpg\">https://www.kaggle.com/vfdev5/noaa-fisheries-steller-sea-lion-population-count/traindotted-train-for-3-7-9-jpg</a></p>",
      "rawMarkdown": "Yes, i also found these 3 images that are not exact copies \n\nhttps://www.kaggle.com/vfdev5/noaa-fisheries-steller-sea-lion-population-count/traindotted-train-for-3-7-9-jpg",
      "votes": null
    },
    {
      "id": "171518",
      "postDate": "03/30/2017 11:41:14",
      "content": "<p>@DataCanary , will the test data be available in kernels ? Thanks for the answer</p>",
      "rawMarkdown": "DataCanary , will the test data be available in kernels ? Thanks for the answer",
      "votes": null
    },
    {
      "id": "171638",
      "postDate": "03/30/2017 21:20:36",
      "content": "<p>Thanks for reporting this! I've posted an <a href=\"https://www.kaggle.com/c/noaa-fisheries-steller-sea-lion-population-count/discussion/30895\">update</a> to the data.</p>",
      "rawMarkdown": "Thanks for reporting this! I've posted an [update](https://www.kaggle.com/c/noaa-fisheries-steller-sea-lion-population-count/discussion/30895) to the data.",
      "votes": null
    },
    {
      "id": "171639",
      "postDate": "03/30/2017 21:21:20",
      "content": "<p>Thanks for the prompt. I've added a Test folder to kernels with the first ten images in it.</p>",
      "rawMarkdown": "Thanks for the prompt. I've added a Test folder to kernels with the first ten images in it.",
      "votes": null
    },
    {
      "id": "172041",
      "postDate": "04/01/2017 17:07:42",
      "content": "<p>I desire coordinates of steller sealions (position x and y). Parsing corrdinates by image processing is not perfect.<br>\nYou use brown and green to indicate place of steller sealions. Brown is color of soil and green is color of plants.</p>",
      "rawMarkdown": "I desire coordinates of steller sealions (position x and y). Parsing corrdinates by image processing is not perfect.<br>\nYou use brown and green to indicate place of steller sealions. Brown is color of soil and green is color of plants.",
      "votes": null
    },
    {
      "id": "172351",
      "postDate": "04/03/2017 12:43:53",
      "content": "<p>I have a question.</p>\n\n<p>In Right Whale Recognition, we made annotations for whale's faces.<br>\n<a href=\"https://www.kaggle.com/c/noaa-right-whale-recognition#creating-a-face-detector-for-whales\">https://www.kaggle.com/c/noaa-right-whale-recognition#creating-a-face-detector-for-whales</a></p>\n\n<p>Is it allowed to make hand annotations in this competition too ?<br></p>",
      "rawMarkdown": "I have a question.\n\nIn Right Whale Recognition, we made annotations for whale's faces.<br>\nhttps://www.kaggle.com/c/noaa-right-whale-recognition#creating-a-face-detector-for-whales\n\nIs it allowed to make hand annotations in this competition too ?<br>",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 170886,
      "author_name": "jiweiliu",
      "author_url": "",
      "post_date": "03/27/2017 22:17:48",
      "content": "<p>Is it possible to have just bounding boxes as annotations in text file? The dotted image is hard to use directly. BBox texts are much smaller too.</p>\n\n<p>for example, <a href=\"http://mscoco.org/dataset/#download\">http://mscoco.org/dataset/#download</a></p>\n\n<p>It really makes no sense to provide duplicated images just for annotations.</p>",
      "votes": null,
      "replies": [
        {
          "id": 170900,
          "author_name": "zerozero",
          "author_url": "",
          "post_date": "03/27/2017 22:52:15",
          "content": "<p>I would second this motion.  Although, I'm guessing that it would be just x,y coordinates per dot.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 170901,
          "author_name": "datacanary",
          "author_url": "",
          "post_date": "03/27/2017 23:02:01",
          "content": "<p>Bounding boxes would be nice, but we don't have the capacity to generate them. I'll be enabling kernels on this competition shortly, though, so people can share their data ingest recipes.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 170958,
          "author_name": "czhang96",
          "author_url": "",
          "post_date": "03/28/2017 05:27:53",
          "content": "<p>I think having a sheet of </p>\n\n<p>train_id, dot_color, x_cord, y_cord</p>\n\n<p>would be greatly helpful. This would make data ingesting much easier, and cut the train set size in half by removing the need of the annotated images.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 171000,
          "author_name": "zfturbo",
          "author_url": "",
          "post_date": "03/28/2017 09:40:14",
          "content": "<p>I double that. We need just text file with dots location. In other case everyone will need to make their own parser of images to find these dots.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 171368,
          "author_name": "mohamedsamy",
          "author_url": "",
          "post_date": "03/29/2017 16:55:39",
          "content": "<p>I have written this notebook to get somehow reliable coordinates of the dots:\n<a href=\"https://www.kaggle.com/mohamedsamy/noaa-fisheries-steller-sea-lion-population-count/generating-text-files-for-dots-coordinates\">https://www.kaggle.com/mohamedsamy/noaa-fisheries-steller-sea-lion-population-count/generating-text-files-for-dots-coordinates</a>\nI have tested it for a couple of images, and I find it pretty good till more robust labels are generated.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 170957,
      "author_name": "czhang96",
      "author_url": "",
      "post_date": "03/28/2017 05:26:11",
      "content": "",
      "votes": null,
      "replies": []
    },
    {
      "id": 171159,
      "author_name": "vfdev5",
      "author_url": "",
      "post_date": "03/28/2017 21:55:45",
      "content": "<p>I wonder why there is no test data available in kernels or I am missing something ?</p>",
      "votes": null,
      "replies": [
        {
          "id": 171518,
          "author_name": "vfdev5",
          "author_url": "",
          "post_date": "03/30/2017 11:41:14",
          "content": "<p>@DataCanary , will the test data be available in kernels ? Thanks for the answer</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 171639,
          "author_name": "datacanary",
          "author_url": "",
          "post_date": "03/30/2017 21:21:20",
          "content": "<p>Thanks for the prompt. I've added a Test folder to kernels with the first ten images in it.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 171403,
      "author_name": "andraszsom",
      "author_url": "",
      "post_date": "03/29/2017 19:49:59",
      "content": "<p>Hi, my understanding is that the Train and TrainDotted images should be the same except for the dots and the black regions in TrainDotted. I'm looking at the small train dataset of 11 images and it seems Train/7.jpg and TrainDotted/7.jpg are mirrored. There is also something weird going on with Train/9.jpg and TrainDotted/9.jpg. They show slightly different areas, so there is no close similarity between the two images. DataCanary, would it be possible to check the other train images and fix these problems?</p>",
      "votes": null,
      "replies": [
        {
          "id": 171513,
          "author_name": "vfdev5",
          "author_url": "",
          "post_date": "03/30/2017 11:24:04",
          "content": "<p>Yes, i also found these 3 images that are not exact copies </p>\n\n<p><a href=\"https://www.kaggle.com/vfdev5/noaa-fisheries-steller-sea-lion-population-count/traindotted-train-for-3-7-9-jpg\">https://www.kaggle.com/vfdev5/noaa-fisheries-steller-sea-lion-population-count/traindotted-train-for-3-7-9-jpg</a></p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 171638,
          "author_name": "datacanary",
          "author_url": "",
          "post_date": "03/30/2017 21:20:36",
          "content": "<p>Thanks for reporting this! I've posted an <a href=\"https://www.kaggle.com/c/noaa-fisheries-steller-sea-lion-population-count/discussion/30895\">update</a> to the data.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 172041,
      "author_name": "toshik",
      "author_url": "",
      "post_date": "04/01/2017 17:07:42",
      "content": "<p>I desire coordinates of steller sealions (position x and y). Parsing corrdinates by image processing is not perfect.<br>\nYou use brown and green to indicate place of steller sealions. Brown is color of soil and green is color of plants.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 172351,
      "author_name": "toshik",
      "author_url": "",
      "post_date": "04/03/2017 12:43:53",
      "content": "<p>I have a question.</p>\n\n<p>In Right Whale Recognition, we made annotations for whale's faces.<br>\n<a href=\"https://www.kaggle.com/c/noaa-right-whale-recognition#creating-a-face-detector-for-whales\">https://www.kaggle.com/c/noaa-right-whale-recognition#creating-a-face-detector-for-whales</a></p>\n\n<p>Is it allowed to make hand annotations in this competition too ?<br></p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "170867": "Welcome everyone to our latest competition! I hope you enjoy it.\n\nBecause this is a large dataset, we'll be making it available via BitTorrent. I'm setting that up now.",
    "170886": "Is it possible to have just bounding boxes as annotations in text file? The dotted image is hard to use directly. BBox texts are much smaller too.\n\nfor example, http://mscoco.org/dataset/#download\n\nIt really makes no sense to provide duplicated images just for annotations.",
    "170900": "I would second this motion.  Although, I'm guessing that it would be just x,y coordinates per dot.",
    "170901": "Bounding boxes would be nice, but we don't have the capacity to generate them. I'll be enabling kernels on this competition shortly, though, so people can share their data ingest recipes.",
    "170957": "",
    "170958": "I think having a sheet of \n\ntrain_id, dot_color, x_cord, y_cord\n \nwould be greatly helpful. This would make data ingesting much easier, and cut the train set size in half by removing the need of the annotated images.",
    "171000": "I double that. We need just text file with dots location. In other case everyone will need to make their own parser of images to find these dots.",
    "171159": "I wonder why there is no test data available in kernels or I am missing something ?",
    "171368": "I have written this notebook to get somehow reliable coordinates of the dots:\nhttps://www.kaggle.com/mohamedsamy/noaa-fisheries-steller-sea-lion-population-count/generating-text-files-for-dots-coordinates\nI have tested it for a couple of images, and I find it pretty good till more robust labels are generated.",
    "171403": "Hi, my understanding is that the Train and TrainDotted images should be the same except for the dots and the black regions in TrainDotted. I'm looking at the small train dataset of 11 images and it seems Train/7.jpg and TrainDotted/7.jpg are mirrored. There is also something weird going on with Train/9.jpg and TrainDotted/9.jpg. They show slightly different areas, so there is no close similarity between the two images. DataCanary, would it be possible to check the other train images and fix these problems?",
    "171513": "Yes, i also found these 3 images that are not exact copies \n\nhttps://www.kaggle.com/vfdev5/noaa-fisheries-steller-sea-lion-population-count/traindotted-train-for-3-7-9-jpg",
    "171518": "DataCanary , will the test data be available in kernels ? Thanks for the answer",
    "171638": "Thanks for reporting this! I've posted an [update](https://www.kaggle.com/c/noaa-fisheries-steller-sea-lion-population-count/discussion/30895) to the data.",
    "171639": "Thanks for the prompt. I've added a Test folder to kernels with the first ten images in it.",
    "172041": "I desire coordinates of steller sealions (position x and y). Parsing corrdinates by image processing is not perfect.<br>\nYou use brown and green to indicate place of steller sealions. Brown is color of soil and green is color of plants.",
    "172351": "I have a question.\n\nIn Right Whale Recognition, we made annotations for whale's faces.<br>\nhttps://www.kaggle.com/c/noaa-right-whale-recognition#creating-a-face-detector-for-whales\n\nIs it allowed to make hand annotations in this competition too ?<br>"
  },
  "source": "meta"
}