{
  "id": 46214,
  "title": "Welcome",
  "url": "/competitions/sp-society-camera-model-identification/discussion/46214",
  "author_name": "",
  "post_date": "2017-12-22T17:10:09.154535700Z",
  "votes": 8,
  "comment_count": 22,
  "views": 0,
  "content": "<p>Welcome to IEEE's Signal Processing Society - Camera Model Identification Challenge!</p>\n\n<p>This is not your run-of-the mill machine learning competition! Be sure to read the Overview, Data, and Evaluation pages carefully. You may also want to check out <a href=\"https://www.kaggle.com/inversion/i-have-no-clue-what-i-m-doing-benchmark\">this kernel</a> to see how to get the correct labels for the training images.</p>\n\n<p>Please feel free to ask your questions in this thread.</p>\n\n<p>Good luck!</p>",
  "messages": [
    {
      "id": "261434",
      "postDate": "12/22/2017 17:10:09",
      "content": "<p>Welcome to IEEE's Signal Processing Society - Camera Model Identification Challenge!</p>\n\n<p>This is not your run-of-the mill machine learning competition! Be sure to read the Overview, Data, and Evaluation pages carefully. You may also want to check out <a href=\"https://www.kaggle.com/inversion/i-have-no-clue-what-i-m-doing-benchmark\">this kernel</a> to see how to get the correct labels for the training images.</p>\n\n<p>Please feel free to ask your questions in this thread.</p>\n\n<p>Good luck!</p>",
      "rawMarkdown": "Welcome to IEEE's Signal Processing Society - Camera Model Identification Challenge!\n\nThis is not your run-of-the mill machine learning competition! Be sure to read the Overview, Data, and Evaluation pages carefully. You may also want to check out [this kernel](https://www.kaggle.com/inversion/i-have-no-clue-what-i-m-doing-benchmark) to see how to get the correct labels for the training images.\n\nPlease feel free to ask your questions in this thread.\n\nGood luck!",
      "votes": null
    },
    {
      "id": "261458",
      "postDate": "12/22/2017 18:19:45",
      "content": "<p>In the Data section this is listed twice:</p>\n\n<blockquote>\n  <p>JPEG compression with quality factor = 90 <br>\n  JPEG compression with quality factor = 90</p>\n</blockquote>\n\n<p>Should there be two different factors?</p>",
      "rawMarkdown": "In the Data section this is listed twice:\n\n&gt; JPEG compression with quality factor = 90  \n&gt; JPEG compression with quality factor = 90\n\nShould there be two different factors?",
      "votes": null
    },
    {
      "id": "261479",
      "postDate": "12/22/2017 19:05:44",
      "content": "<p>Thanks . . . the first factor should be 70. Page updated!</p>",
      "rawMarkdown": "Thanks . . . the first factor should be 70. Page updated!",
      "votes": null
    },
    {
      "id": "261525",
      "postDate": "12/22/2017 20:43:01",
      "content": "<p>Will a manipulated image only have 1 of the possible processing operations applied it or can an image potentially have multiple operations applied to it?</p>",
      "rawMarkdown": "Will a manipulated image only have 1 of the possible processing operations applied it or can an image potentially have multiple operations applied to it?",
      "votes": null
    },
    {
      "id": "261526",
      "postDate": "12/22/2017 20:46:56",
      "content": "<p>Quick question: how many devices does Ben Hamner have?</p>",
      "rawMarkdown": "Quick question: how many devices does Ben Hamner have?",
      "votes": null
    },
    {
      "id": "261536",
      "postDate": "12/22/2017 21:33:52",
      "content": "<p>Only one at a time. ;-)</p>",
      "rawMarkdown": "Only one at a time. ;-)",
      "votes": null
    },
    {
      "id": "261537",
      "postDate": "12/22/2017 21:34:36",
      "content": "<p>Just one.</p>",
      "rawMarkdown": "Just one.",
      "votes": null
    },
    {
      "id": "261971",
      "postDate": "12/24/2017 18:07:59",
      "content": "<p>Were all the images from the train set and for a specific model taken with the same device ? </p>",
      "rawMarkdown": "Were all the images from the train set and for a specific model taken with the same device ?",
      "votes": null
    },
    {
      "id": "262057",
      "postDate": "12/25/2017 05:12:58",
      "content": "<p>Deadlines  mentioned in the SP-Cup document\nJanuary 21, 2018 Deadline   for submitting  open    competition results\nFebruary    10, 2018 Announcement   of  the three   finalist    teams</p>\n\n<p>But in the Competition timeline \nFebruary 1, 2018 - Entry deadline. You must accept the competition rules before this date in order to compete.\nFebruary 1, 2018 - Team Merger deadline. This is the last day participants may join or merge teams.\nFebruary 8, 2018 - Final submission deadline.</p>\n\n<p>Which one is the correct one for eligible teams?</p>",
      "rawMarkdown": "Deadlines  mentioned in the SP-Cup document\nJanuary\t21,\t2018 Deadline\tfor\tsubmitting\topen\tcompetition\tresults\nFebruary\t10,\t2018 Announcement\tof\tthe\tthree\tfinalist\tteams\n\nBut in the Competition timeline \nFebruary 1, 2018 - Entry deadline. You must accept the competition rules before this date in order to compete.\nFebruary 1, 2018 - Team Merger deadline. This is the last day participants may join or merge teams.\nFebruary 8, 2018 - Final submission deadline.\n\nWhich one is the correct one for eligible teams?",
      "votes": null
    },
    {
      "id": "262266",
      "postDate": "12/26/2017 02:32:39",
      "content": "<p>Hi Marwan,</p>\n\n<p>I will confirm with the competition sponsor on the difference in the deadlines between the document and the Kaggle competitions. </p>",
      "rawMarkdown": "Hi Marwan,\n\nI will confirm with the competition sponsor on the difference in the deadlines between the document and the Kaggle competitions.",
      "votes": null
    },
    {
      "id": "262959",
      "postDate": "12/28/2017 12:47:44",
      "content": "<p>emmm, a question I want to inquire  is that  can we use the dataset, like the Dresden image database, to train a model as pre-train model?</p>\n\n<p>Best</p>",
      "rawMarkdown": "emmm, a question I want to inquire  is that  can we use the dataset, like the Dresden image database, to train a model as pre-train model?\n\nBest",
      "votes": null
    },
    {
      "id": "264422",
      "postDate": "01/03/2018 02:23:45",
      "content": "<p>The evaluation function may be $$\\sum_{i=1}^{n}\\frac{w_i(y_i=\\hat{y}_i)}{\\sum w_j},$$ not  $$\\frac{1}{n}\\sum_{i=1}^{n}\\frac{w_i(y_i=\\hat{y}_i)}{\\sum w_j},$$</p>\n\n<p>since for the latter the maximum score will never be equal to 1.</p>",
      "rawMarkdown": "The evaluation function may be $$\\sum_{i=1}^{n}\\frac{w_i(y_i=\\hat{y}_i)}{\\sum w_j},$$ not  $$\\frac{1}{n}\\sum_{i=1}^{n}\\frac{w_i(y_i=\\hat{y}_i)}{\\sum w_j},$$\n\nsince for the latter the maximum score will never be equal to 1.",
      "votes": null
    },
    {
      "id": "265783",
      "postDate": "01/06/2018 17:15:33",
      "content": "<p>Are images cropped to 512x512 before or after manipulation? Thanks!</p>",
      "rawMarkdown": "Are images cropped to 512x512 before or after manipulation? Thanks!",
      "votes": null
    },
    {
      "id": "265844",
      "postDate": "01/06/2018 21:00:16",
      "content": "<p>Hi there. Is is allowable to use external data to get more training data?</p>",
      "rawMarkdown": "Hi there. Is is allowable to use external data to get more training data?",
      "votes": null
    },
    {
      "id": "266448",
      "postDate": "01/08/2018 19:21:17",
      "content": "<p>I have a naive question, Is it allowed to use the information contained in the test image file names to process differently altered and original images ?\nThanks,\nJean</p>",
      "rawMarkdown": "I have a naive question, Is it allowed to use the information contained in the test image file names to process differently altered and original images ?\nThanks,\nJean",
      "votes": null
    },
    {
      "id": "269191",
      "postDate": "01/16/2018 12:33:43",
      "content": "<p>Do you allow that we use this dataset in that competition? <br>\n<a href=\"https://www.kaggle.com/c/sp-society-camera-model-identification/discussion/47235\">https://www.kaggle.com/c/sp-society-camera-model-identification/discussion/47235</a></p>",
      "rawMarkdown": "Do you allow that we use this dataset in that competition?  \nhttps://www.kaggle.com/c/sp-society-camera-model-identification/discussion/47235",
      "votes": null
    },
    {
      "id": "271480",
      "postDate": "01/20/2018 14:57:43",
      "content": "<p>Hi, Question:</p>\n\n<p>Can you elaborate how the resizing manipulations are done, e.g:</p>\n\n<p>resizing (via bicubic interpolation) by a factor of 0.5</p>\n\n<p>Does it mean that the crop taken from the initial image was 1024x1024 and the resulting manipulated image was resized down to 512x512 (0.5x)?</p>",
      "rawMarkdown": "Hi, Question:\n\nCan you elaborate how the resizing manipulations are done, e.g:\n\nresizing (via bicubic interpolation) by a factor of 0.5\n\nDoes it mean that the crop taken from the initial image was 1024x1024 and the resulting manipulated image was resized down to 512x512 (0.5x)?",
      "votes": null
    },
    {
      "id": "271549",
      "postDate": "01/20/2018 18:42:34",
      "content": "<p>Would be nice to have a response on this to avoid a Favorita 2.0. Per competition rules:</p>\n\n<p>\"External Data Use: The following provision supersedes General Rules Section 7.B. below: “You may use data, other than the Competition Data, as allowed on the Competition Website to develop and test your models and Submissions; provided, you have the right and authority to use such external data for the purposes of the Competition, and to share such data with Sponsor and Kaggle as may be required.\"</p>\n\n<p>Based on the above and the fact it seems that the Flickr images above have a creative commons license that allows users to :\nShare — copy and redistribute the material in any medium or format\nAdapt — remix, transform, and build upon the material\nfor any purpose, even commercially. (all of the ones that I checked so far, and I picked out 30) I think that it is acceptable to use for the competition,  but I'm not a lawyer. </p>",
      "rawMarkdown": "Would be nice to have a response on this to avoid a Favorita 2.0. Per competition rules:\n\n \"External Data Use: The following provision supersedes General Rules Section 7.B. below: “You may use data, other than the Competition Data, as allowed on the Competition Website to develop and test your models and Submissions; provided, you have the right and authority to use such external data for the purposes of the Competition, and to share such data with Sponsor and Kaggle as may be required.\"\n\nBased on the above and the fact it seems that the Flickr images above have a creative commons license that allows users to :\nShare — copy and redistribute the material in any medium or format\nAdapt — remix, transform, and build upon the material\nfor any purpose, even commercially. (all of the ones that I checked so far, and I picked out 30) I think that it is acceptable to use for the competition,  but I'm not a lawyer.",
      "votes": null
    },
    {
      "id": "271602",
      "postDate": "01/20/2018 23:06:26",
      "content": "<p>Quite a few of the photos are CC-Non-Commercial which makes things murky. \n(<a href=\"https://creativecommons.org/licenses/by-nc/2.0/\">https://creativecommons.org/licenses/by-nc/2.0/</a>)\n\"NonCommercial — You may not use the material for commercial purposes.\"</p>\n\n<p>e.g., under htc_m7\nalbum link: <a href=\"https://www.flickr.com/photos/153787453@N07/36435746644/\">https://www.flickr.com/photos/153787453@N07/36435746644/</a>\ndirect link: <a href=\"https://c1.staticflickr.com/5/4431/36435746644_7fdd350c9a_o.jpg\">https://c1.staticflickr.com/5/4431/36435746644_7fdd350c9a_o.jpg</a></p>\n\n<p>But yeah, would be nice to hear from the organizers about this. </p>",
      "rawMarkdown": "Quite a few of the photos are CC-Non-Commercial which makes things murky. \n(https://creativecommons.org/licenses/by-nc/2.0/)\n\"NonCommercial — You may not use the material for commercial purposes.\"\n\ne.g., under htc_m7\nalbum link: https://www.flickr.com/photos/153787453@N07/36435746644/\ndirect link: https://c1.staticflickr.com/5/4431/36435746644_7fdd350c9a_o.jpg\n\nBut yeah, would be nice to hear from the organizers about this.",
      "votes": null
    },
    {
      "id": "271604",
      "postDate": "01/20/2018 23:16:18",
      "content": "<p>You're right. Thanks for pointing that out. </p>\n\n<p>I guess then it comes down to whether the competition is \"commercial\"== \"primarily for profit\". Is that based on IEEE's status? I don't know... </p>\n\n<p>It also matters in terms of score. An identical model trained with and without the extra data scored .918 vs .904. So a boost of .014, which is not trivial </p>",
      "rawMarkdown": "You're right. Thanks for pointing that out. \n\nI guess then it comes down to whether the competition is \"commercial\"== \"primarily for profit\". Is that based on IEEE's status? I don't know... \n\nIt also matters in terms of score. An identical model trained with and without the extra data scored .918 vs .904. So a boost of .014, which is not trivial",
      "votes": null
    },
    {
      "id": "271617",
      "postDate": "01/20/2018 23:52:06",
      "content": "<p>Yeah, agreed.  </p>\n\n<p>Oddly enough, a quick Google search brings up the fact that ImageNet itself is built on copyrighted images and usage of it is for research and non-commercial use only. How many startups are built on pre-trained ImageNet models...?</p>",
      "rawMarkdown": "Yeah, agreed.  \n\nOddly enough, a quick Google search brings up the fact that ImageNet itself is built on copyrighted images and usage of it is for research and non-commercial use only. How many startups are built on pre-trained ImageNet models...?",
      "votes": null
    },
    {
      "id": "271619",
      "postDate": "01/21/2018 00:06:35",
      "content": "<p>That's pretty funny. Plus pre-trained models are used all of the time for Kaggle competitions as well...</p>",
      "rawMarkdown": "That's pretty funny. Plus pre-trained models are used all of the time for Kaggle competitions as well...",
      "votes": null
    },
    {
      "id": "276558",
      "postDate": "01/31/2018 18:42:55",
      "content": "<p>Is the test set for current leader board based on a random mixture of camera models? I sometimes see a big difference between my local CV and leader board, and note that my model does better for same camera models than others. Thanks</p>",
      "rawMarkdown": "Is the test set for current leader board based on a random mixture of camera models? I sometimes see a big difference between my local CV and leader board, and note that my model does better for same camera models than others. Thanks",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 261458,
      "author_name": "brandenkmurray",
      "author_url": "",
      "post_date": "12/22/2017 18:19:45",
      "content": "<p>In the Data section this is listed twice:</p>\n\n<blockquote>\n  <p>JPEG compression with quality factor = 90 <br>\n  JPEG compression with quality factor = 90</p>\n</blockquote>\n\n<p>Should there be two different factors?</p>",
      "votes": null,
      "replies": [
        {
          "id": 261479,
          "author_name": "inversion",
          "author_url": "",
          "post_date": "12/22/2017 19:05:44",
          "content": "<p>Thanks . . . the first factor should be 70. Page updated!</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 261525,
      "author_name": "brandenkmurray",
      "author_url": "",
      "post_date": "12/22/2017 20:43:01",
      "content": "<p>Will a manipulated image only have 1 of the possible processing operations applied it or can an image potentially have multiple operations applied to it?</p>",
      "votes": null,
      "replies": [
        {
          "id": 261537,
          "author_name": "inversion",
          "author_url": "",
          "post_date": "12/22/2017 21:34:36",
          "content": "<p>Just one.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 261526,
      "author_name": "tunguz",
      "author_url": "",
      "post_date": "12/22/2017 20:46:56",
      "content": "<p>Quick question: how many devices does Ben Hamner have?</p>",
      "votes": null,
      "replies": [
        {
          "id": 261536,
          "author_name": "inversion",
          "author_url": "",
          "post_date": "12/22/2017 21:33:52",
          "content": "<p>Only one at a time. ;-)</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 261971,
      "author_name": "CVxTz",
      "author_url": "",
      "post_date": "12/24/2017 18:07:59",
      "content": "<p>Were all the images from the train set and for a specific model taken with the same device ? </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 262057,
      "author_name": "marwantorki",
      "author_url": "",
      "post_date": "12/25/2017 05:12:58",
      "content": "<p>Deadlines  mentioned in the SP-Cup document\nJanuary 21, 2018 Deadline   for submitting  open    competition results\nFebruary    10, 2018 Announcement   of  the three   finalist    teams</p>\n\n<p>But in the Competition timeline \nFebruary 1, 2018 - Entry deadline. You must accept the competition rules before this date in order to compete.\nFebruary 1, 2018 - Team Merger deadline. This is the last day participants may join or merge teams.\nFebruary 8, 2018 - Final submission deadline.</p>\n\n<p>Which one is the correct one for eligible teams?</p>",
      "votes": null,
      "replies": [
        {
          "id": 262266,
          "author_name": "addisonhoward",
          "author_url": "",
          "post_date": "12/26/2017 02:32:39",
          "content": "<p>Hi Marwan,</p>\n\n<p>I will confirm with the competition sponsor on the difference in the deadlines between the document and the Kaggle competitions. </p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 262959,
      "author_name": "oujiayu",
      "author_url": "",
      "post_date": "12/28/2017 12:47:44",
      "content": "<p>emmm, a question I want to inquire  is that  can we use the dataset, like the Dresden image database, to train a model as pre-train model?</p>\n\n<p>Best</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 264422,
      "author_name": "usingtc",
      "author_url": "",
      "post_date": "01/03/2018 02:23:45",
      "content": "<p>The evaluation function may be $$\\sum_{i=1}^{n}\\frac{w_i(y_i=\\hat{y}_i)}{\\sum w_j},$$ not  $$\\frac{1}{n}\\sum_{i=1}^{n}\\frac{w_i(y_i=\\hat{y}_i)}{\\sum w_j},$$</p>\n\n<p>since for the latter the maximum score will never be equal to 1.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 265783,
      "author_name": "maltonji",
      "author_url": "",
      "post_date": "01/06/2018 17:15:33",
      "content": "<p>Are images cropped to 512x512 before or after manipulation? Thanks!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 265844,
      "author_name": "xiaokangwang",
      "author_url": "",
      "post_date": "01/06/2018 21:00:16",
      "content": "<p>Hi there. Is is allowable to use external data to get more training data?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 266448,
      "author_name": "jeandebleau",
      "author_url": "",
      "post_date": "01/08/2018 19:21:17",
      "content": "<p>I have a naive question, Is it allowed to use the information contained in the test image file names to process differently altered and original images ?\nThanks,\nJean</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 269191,
      "author_name": "tereka",
      "author_url": "",
      "post_date": "01/16/2018 12:33:43",
      "content": "<p>Do you allow that we use this dataset in that competition? <br>\n<a href=\"https://www.kaggle.com/c/sp-society-camera-model-identification/discussion/47235\">https://www.kaggle.com/c/sp-society-camera-model-identification/discussion/47235</a></p>",
      "votes": null,
      "replies": [
        {
          "id": 271549,
          "author_name": "rhgrossm",
          "author_url": "",
          "post_date": "01/20/2018 18:42:34",
          "content": "<p>Would be nice to have a response on this to avoid a Favorita 2.0. Per competition rules:</p>\n\n<p>\"External Data Use: The following provision supersedes General Rules Section 7.B. below: “You may use data, other than the Competition Data, as allowed on the Competition Website to develop and test your models and Submissions; provided, you have the right and authority to use such external data for the purposes of the Competition, and to share such data with Sponsor and Kaggle as may be required.\"</p>\n\n<p>Based on the above and the fact it seems that the Flickr images above have a creative commons license that allows users to :\nShare — copy and redistribute the material in any medium or format\nAdapt — remix, transform, and build upon the material\nfor any purpose, even commercially. (all of the ones that I checked so far, and I picked out 30) I think that it is acceptable to use for the competition,  but I'm not a lawyer. </p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 271602,
          "author_name": "leecming",
          "author_url": "",
          "post_date": "01/20/2018 23:06:26",
          "content": "<p>Quite a few of the photos are CC-Non-Commercial which makes things murky. \n(<a href=\"https://creativecommons.org/licenses/by-nc/2.0/\">https://creativecommons.org/licenses/by-nc/2.0/</a>)\n\"NonCommercial — You may not use the material for commercial purposes.\"</p>\n\n<p>e.g., under htc_m7\nalbum link: <a href=\"https://www.flickr.com/photos/153787453@N07/36435746644/\">https://www.flickr.com/photos/153787453@N07/36435746644/</a>\ndirect link: <a href=\"https://c1.staticflickr.com/5/4431/36435746644_7fdd350c9a_o.jpg\">https://c1.staticflickr.com/5/4431/36435746644_7fdd350c9a_o.jpg</a></p>\n\n<p>But yeah, would be nice to hear from the organizers about this. </p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 271604,
          "author_name": "rhgrossm",
          "author_url": "",
          "post_date": "01/20/2018 23:16:18",
          "content": "<p>You're right. Thanks for pointing that out. </p>\n\n<p>I guess then it comes down to whether the competition is \"commercial\"== \"primarily for profit\". Is that based on IEEE's status? I don't know... </p>\n\n<p>It also matters in terms of score. An identical model trained with and without the extra data scored .918 vs .904. So a boost of .014, which is not trivial </p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 271617,
          "author_name": "leecming",
          "author_url": "",
          "post_date": "01/20/2018 23:52:06",
          "content": "<p>Yeah, agreed.  </p>\n\n<p>Oddly enough, a quick Google search brings up the fact that ImageNet itself is built on copyrighted images and usage of it is for research and non-commercial use only. How many startups are built on pre-trained ImageNet models...?</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 271619,
          "author_name": "rhgrossm",
          "author_url": "",
          "post_date": "01/21/2018 00:06:35",
          "content": "<p>That's pretty funny. Plus pre-trained models are used all of the time for Kaggle competitions as well...</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 271480,
      "author_name": "antorsae",
      "author_url": "",
      "post_date": "01/20/2018 14:57:43",
      "content": "<p>Hi, Question:</p>\n\n<p>Can you elaborate how the resizing manipulations are done, e.g:</p>\n\n<p>resizing (via bicubic interpolation) by a factor of 0.5</p>\n\n<p>Does it mean that the crop taken from the initial image was 1024x1024 and the resulting manipulated image was resized down to 512x512 (0.5x)?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 276558,
      "author_name": "bgeier",
      "author_url": "",
      "post_date": "01/31/2018 18:42:55",
      "content": "<p>Is the test set for current leader board based on a random mixture of camera models? I sometimes see a big difference between my local CV and leader board, and note that my model does better for same camera models than others. Thanks</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "261434": "Welcome to IEEE's Signal Processing Society - Camera Model Identification Challenge!\n\nThis is not your run-of-the mill machine learning competition! Be sure to read the Overview, Data, and Evaluation pages carefully. You may also want to check out [this kernel](https://www.kaggle.com/inversion/i-have-no-clue-what-i-m-doing-benchmark) to see how to get the correct labels for the training images.\n\nPlease feel free to ask your questions in this thread.\n\nGood luck!",
    "261458": "In the Data section this is listed twice:\n\n&gt; JPEG compression with quality factor = 90  \n&gt; JPEG compression with quality factor = 90\n\nShould there be two different factors?",
    "261479": "Thanks . . . the first factor should be 70. Page updated!",
    "261525": "Will a manipulated image only have 1 of the possible processing operations applied it or can an image potentially have multiple operations applied to it?",
    "261526": "Quick question: how many devices does Ben Hamner have?",
    "261536": "Only one at a time. ;-)",
    "261537": "Just one.",
    "261971": "Were all the images from the train set and for a specific model taken with the same device ?",
    "262057": "Deadlines  mentioned in the SP-Cup document\nJanuary\t21,\t2018 Deadline\tfor\tsubmitting\topen\tcompetition\tresults\nFebruary\t10,\t2018 Announcement\tof\tthe\tthree\tfinalist\tteams\n\nBut in the Competition timeline \nFebruary 1, 2018 - Entry deadline. You must accept the competition rules before this date in order to compete.\nFebruary 1, 2018 - Team Merger deadline. This is the last day participants may join or merge teams.\nFebruary 8, 2018 - Final submission deadline.\n\nWhich one is the correct one for eligible teams?",
    "262266": "Hi Marwan,\n\nI will confirm with the competition sponsor on the difference in the deadlines between the document and the Kaggle competitions.",
    "262959": "emmm, a question I want to inquire  is that  can we use the dataset, like the Dresden image database, to train a model as pre-train model?\n\nBest",
    "264422": "The evaluation function may be $$\\sum_{i=1}^{n}\\frac{w_i(y_i=\\hat{y}_i)}{\\sum w_j},$$ not  $$\\frac{1}{n}\\sum_{i=1}^{n}\\frac{w_i(y_i=\\hat{y}_i)}{\\sum w_j},$$\n\nsince for the latter the maximum score will never be equal to 1.",
    "265783": "Are images cropped to 512x512 before or after manipulation? Thanks!",
    "265844": "Hi there. Is is allowable to use external data to get more training data?",
    "266448": "I have a naive question, Is it allowed to use the information contained in the test image file names to process differently altered and original images ?\nThanks,\nJean",
    "269191": "Do you allow that we use this dataset in that competition?  \nhttps://www.kaggle.com/c/sp-society-camera-model-identification/discussion/47235",
    "271480": "Hi, Question:\n\nCan you elaborate how the resizing manipulations are done, e.g:\n\nresizing (via bicubic interpolation) by a factor of 0.5\n\nDoes it mean that the crop taken from the initial image was 1024x1024 and the resulting manipulated image was resized down to 512x512 (0.5x)?",
    "271549": "Would be nice to have a response on this to avoid a Favorita 2.0. Per competition rules:\n\n \"External Data Use: The following provision supersedes General Rules Section 7.B. below: “You may use data, other than the Competition Data, as allowed on the Competition Website to develop and test your models and Submissions; provided, you have the right and authority to use such external data for the purposes of the Competition, and to share such data with Sponsor and Kaggle as may be required.\"\n\nBased on the above and the fact it seems that the Flickr images above have a creative commons license that allows users to :\nShare — copy and redistribute the material in any medium or format\nAdapt — remix, transform, and build upon the material\nfor any purpose, even commercially. (all of the ones that I checked so far, and I picked out 30) I think that it is acceptable to use for the competition,  but I'm not a lawyer.",
    "271602": "Quite a few of the photos are CC-Non-Commercial which makes things murky. \n(https://creativecommons.org/licenses/by-nc/2.0/)\n\"NonCommercial — You may not use the material for commercial purposes.\"\n\ne.g., under htc_m7\nalbum link: https://www.flickr.com/photos/153787453@N07/36435746644/\ndirect link: https://c1.staticflickr.com/5/4431/36435746644_7fdd350c9a_o.jpg\n\nBut yeah, would be nice to hear from the organizers about this.",
    "271604": "You're right. Thanks for pointing that out. \n\nI guess then it comes down to whether the competition is \"commercial\"== \"primarily for profit\". Is that based on IEEE's status? I don't know... \n\nIt also matters in terms of score. An identical model trained with and without the extra data scored .918 vs .904. So a boost of .014, which is not trivial",
    "271617": "Yeah, agreed.  \n\nOddly enough, a quick Google search brings up the fact that ImageNet itself is built on copyrighted images and usage of it is for research and non-commercial use only. How many startups are built on pre-trained ImageNet models...?",
    "271619": "That's pretty funny. Plus pre-trained models are used all of the time for Kaggle competitions as well...",
    "276558": "Is the test set for current leader board based on a random mixture of camera models? I sometimes see a big difference between my local CV and leader board, and note that my model does better for same camera models than others. Thanks"
  },
  "source": "meta"
}