{
  "id": 70005,
  "title": "Reminder: MODEL UPLOAD DEADLINE",
  "url": "/competitions/inclusive-images-challenge/discussion/70005",
  "author_name": "Julia Elliott",
  "post_date": "2018-10-29T20:40:55.133000",
  "votes": 5,
  "comment_count": 6,
  "views": 0,
  "content": "<p>This is your reminder that November 5th (next week) is the close of stage 1 and deadline to upload your model. Model upload can be completed by navigating to the <a href=\"https://www.kaggle.com/c/inclusive-images-challenge/team\">Team</a> tab, and scrolling down to \"Your Model.\" There will be no exceptions to this deadline, and prize eligibility will be contingent on compliance with this requirement.</p>\n\n<p>Upon the close of stage 1, stage 2 data will be released on November 6th, with only 6 days to submit your predictions on that test set. Note that you will only be permitted to select 1 final submission for scoring. This selected submission is what will be expected to match the uploaded model. See the <a href=\"https://www.kaggle.com/c/inclusive-images-challenge#Timeline\">timeline</a> page for full details, noting that all deadlines are at 11:59 PM UTC on the corresponding day.</p>\n\n<p>Please refer to Kaggle's <a href=\"https://www.kaggle.com/two-stage-frequently-asked-questions\">2-stage competition FAQ</a> for more information about the 2-stage design. However, note the following important distinctions:</p>\n\n<ul>\n<li>In this 2-stage competition, you are <strong>not permitted</strong> to retrain your models after stage 1. Nor are you permitted to use any data for training, other than <a href=\"https://www.kaggle.com/c/inclusive-images-challenge#Data-Download-&amp;-Getting-Started\">the dataset specified as the eligible train set for this competition</a>. This is an explicit difference from the 2-stage FAQ linked above.</li>\n<li>It should be noted that the stage 2 test set will contain images that are blurred but do not contain people. This is in line with the <a href=\"https://www.kaggle.com/c/inclusive-images-challenge#Inclusive-Images-FAQ\">Competition FAQ</a> specifying that competitors should not use dataset peculiarities in their modeling methods.</li>\n<li>Please ensure compliance with the <a href=\"https://www.kaggle.com/c/inclusive-images-challenge#Model-Eligibility-Requirements\">Model Eligibility Requirements</a>.</li>\n</ul>\n\n<p>Best of luck!</p>",
  "messages": [
    {
      "id": 412239,
      "postDate": "2018-10-29T20:40:55.133Z",
      "content": "<p>This is your reminder that November 5th (next week) is the close of stage 1 and deadline to upload your model. Model upload can be completed by navigating to the <a href=\"https://www.kaggle.com/c/inclusive-images-challenge/team\">Team</a> tab, and scrolling down to \"Your Model.\" There will be no exceptions to this deadline, and prize eligibility will be contingent on compliance with this requirement.</p>\n\n<p>Upon the close of stage 1, stage 2 data will be released on November 6th, with only 6 days to submit your predictions on that test set. Note that you will only be permitted to select 1 final submission for scoring. This selected submission is what will be expected to match the uploaded model. See the <a href=\"https://www.kaggle.com/c/inclusive-images-challenge#Timeline\">timeline</a> page for full details, noting that all deadlines are at 11:59 PM UTC on the corresponding day.</p>\n\n<p>Please refer to Kaggle's <a href=\"https://www.kaggle.com/two-stage-frequently-asked-questions\">2-stage competition FAQ</a> for more information about the 2-stage design. However, note the following important distinctions:</p>\n\n<ul>\n<li>In this 2-stage competition, you are <strong>not permitted</strong> to retrain your models after stage 1. Nor are you permitted to use any data for training, other than <a href=\"https://www.kaggle.com/c/inclusive-images-challenge#Data-Download-&amp;-Getting-Started\">the dataset specified as the eligible train set for this competition</a>. This is an explicit difference from the 2-stage FAQ linked above.</li>\n<li>It should be noted that the stage 2 test set will contain images that are blurred but do not contain people. This is in line with the <a href=\"https://www.kaggle.com/c/inclusive-images-challenge#Inclusive-Images-FAQ\">Competition FAQ</a> specifying that competitors should not use dataset peculiarities in their modeling methods.</li>\n<li>Please ensure compliance with the <a href=\"https://www.kaggle.com/c/inclusive-images-challenge#Model-Eligibility-Requirements\">Model Eligibility Requirements</a>.</li>\n</ul>\n\n<p>Best of luck!</p>",
      "rawMarkdown": "This is your reminder that November 5th (next week) is the close of stage 1 and deadline to upload your model. Model upload can be completed by navigating to the [Team][1] tab, and scrolling down to \"Your Model.\" There will be no exceptions to this deadline, and prize eligibility will be contingent on compliance with this requirement.\n\nUpon the close of stage 1, stage 2 data will be released on November 6th, with only 6 days to submit your predictions on that test set. Note that you will only be permitted to select 1 final submission for scoring. This selected submission is what will be expected to match the uploaded model. See the [timeline][2] page for full details, noting that all deadlines are at 11:59 PM UTC on the corresponding day.\n\nPlease refer to Kaggle's [2-stage competition FAQ][3] for more information about the 2-stage design. However, note the following important distinctions:\n\n - In this 2-stage competition, you are **not permitted** to retrain your models after stage 1. Nor are you permitted to use any data for training, other than [the dataset specified as the eligible train set for this competition][4]. This is an explicit difference from the 2-stage FAQ linked above.\n - It should be noted that the stage 2 test set will contain images that are blurred but do not contain people. This is in line with the [Competition FAQ][5] specifying that competitors should not use dataset peculiarities in their modeling methods.\n - Please ensure compliance with the [Model Eligibility Requirements][6].\n\nBest of luck!\n\n\n  [1]: https://www.kaggle.com/c/inclusive-images-challenge/team\n  [2]: https://www.kaggle.com/c/inclusive-images-challenge#Timeline\n  [3]: https://www.kaggle.com/two-stage-frequently-asked-questions\n  [4]: https://www.kaggle.com/c/inclusive-images-challenge#Data-Download-&amp;-Getting-Started\n  [5]: https://www.kaggle.com/c/inclusive-images-challenge#Inclusive-Images-FAQ\n  [6]: https://www.kaggle.com/c/inclusive-images-challenge#Model-Eligibility-Requirements",
      "votes": 4
    },
    {
      "id": 414888,
      "postDate": "2018-11-03T20:45:05.330Z",
      "content": "<p>Can we upload multiple checkpoints of our model, which basically mean multiple models? Since it is more convenient to make ensembling.</p>\n\n<p>And the code should include stage 1 inference code, and also that for stage 2?</p>",
      "rawMarkdown": "Can we upload multiple checkpoints of our model, which basically mean multiple models? Since it is more convenient to make ensembling.\n\nAnd the code should include stage 1 inference code, and also that for stage 2?",
      "votes": 1
    },
    {
      "id": 412630,
      "postDate": "2018-10-30T14:22:34.373Z",
      "content": "<p>Does it mean that we need to upload a script (for example, python script) to generate valid submission file and model weights (for example, *.hdf5 files with weights) and exactly those weights and script must be used to generate predictions for stage 2 ?</p>",
      "rawMarkdown": "Does it mean that we need to upload a script (for example, python script) to generate valid submission file and model weights (for example, *.hdf5 files with weights) and exactly those weights and script must be used to generate predictions for stage 2 ?",
      "votes": 1,
      "replies": [
        {
          "id": 412661,
          "postDate": "2018-10-30T15:19:18.990Z",
          "content": "<p>That's correct.</p>",
          "rawMarkdown": "That's correct.",
          "votes": 2
        }
      ]
    },
    {
      "id": 415438,
      "postDate": "2018-11-05T05:45:16.287Z",
      "content": "<p>does the file <a href=\"https://storage.googleapis.com/openimages/2018_04/image_ids_and_rotation.csv\">image_ids_and_rotation.csv</a> contain the information about correlation between image names and their id's?\nIs there a sane way to get images sorted to classes in a form of sub-folder structure as for caffe model training?\nI am asking, because what I can see with the</p>\n\n<pre><code> cat image_ids_and_rotation.csv | tail\n</code></pre>\n\n<p>is the information below from where idk how to determine what is the image id:</p>\n\n<p>&gt; <a href=\"https://farm6.staticflickr.com/3835/15022727040_f6b680d8bc_o.jpg,https://www.flickr.com/photos/davehamster/15022727040,https://creativecommons.org/licenses/by/2.0/,https://www.flickr.com/people/davehamster/,David\">https://farm6.staticflickr.com/3835/15022727040_f6b680d8bc_o.jpg,https://www.flickr.com/photos/davehamster/15022727040,https://creativecommons.org/licenses/by/2.0/,https://www.flickr.com/people/davehamster/,David</a>\n&gt; Merrett,1934 Riley TT\n&gt; Sprite,745972,TYaYy6M4D7qP7X7ZcPc1uw==,<a href=\"https://c2.staticflickr.com/4/3835/15022727040_818e71880c_z.jpg,0\">https://c2.staticflickr.com/4/3835/15022727040_818e71880c_z.jpg,0</a>\n&gt; 8a23c798ea665d15,train,<a href=\"https://farm6.staticflickr.com/7180/6973157007_6d633ffef3_o.jpg,https://www.flickr.com/photos/mejiaperalta/6973157007,https://creativecommons.org/licenses/by/2.0/,https://www.flickr.com/people/mejiaperalta/,Jorge\">https://farm6.staticflickr.com/7180/6973157007_6d633ffef3_o.jpg,https://www.flickr.com/photos/mejiaperalta/6973157007,https://creativecommons.org/licenses/by/2.0/,https://www.flickr.com/people/mejiaperalta/,Jorge</a>\n&gt; Mejía\n&gt; \n&gt; peralta,IMG_2936tepe,1214382,qaXSvOxNbkjx6hUOeoYDJg==,<a href=\"https://c4.staticflickr.com/8/7180/6973157007_0cc4687da9_z.jpg,0\">https://c4.staticflickr.com/8/7180/6973157007_0cc4687da9_z.jpg,0</a>\n&gt; 90d0031f89de28ee,train,<a href=\"https://c7.staticflickr.com/1/23/27246359_825538fa3e_o.jpg,https://www.flickr.com/photos/emdot/27246359,https://creativecommons.org/licenses/by/2.0/,https://www.flickr.com/people/emdot/,emdot,i\">https://c7.staticflickr.com/1/23/27246359_825538fa3e_o.jpg,https://www.flickr.com/photos/emdot/27246359,https://creativecommons.org/licenses/by/2.0/,https://www.flickr.com/people/emdot/,emdot,i</a>\n&gt; swear it was there just a minute\n&gt; ago,561560,i6YWnWUyYlkjTCiOCoQhHA==,<a href=\"https://c6.staticflickr.com/1/23/27246359_825538fa3e_z.jpg?zz=1,0\">https://c6.staticflickr.com/1/23/27246359_825538fa3e_z.jpg?zz=1,0</a></p>\n\n<p>What is the image id in the excerpt above?</p>\n\n<p>upon checking it appears that images id's are aither in the file <a href=\"https://storage.googleapis.com/openimages/2018_04/train/train-annotations-human-imagelabels-boxable.csv\">https://storage.googleapis.com/openimages/2018_04/train/train-annotations-human-imagelabels-boxable.csv</a> or in the file <a href=\"https://storage.googleapis.com/openimages/2018_04/train/train-images-boxable-with-rotation.csv\">https://storage.googleapis.com/openimages/2018_04/train/train-images-boxable-with-rotation.csv</a>\nThank you.\nImages id's have been located:</p>\n\n<pre><code>cat train-annotations-human-imagelabels-boxable.csv | tail\n</code></pre>\n\n<p>ffffeac7d2f37489,verification,/m/07yv9,1\nffffeac7d2f37489,verification,/m/083wq,1\nffffeac7d2f37489,verification,/m/0h9mv,1\nffffeac7d2f37489,verification,/m/0k4j,1\nffffebddbb8bba21,verification,/m/0242l,1\nfffffdaec951185d,verification,/m/01g317,1\nfffffdaec951185d,verification,/m/01xyhv,1\nfffffdaec951185d,verification,/m/03bt1vf,1\nfffffdaec951185d,verification,/m/09j2d,1\nfffffdaec951185d,verification,/m/0dzct,1</p>\n\n<p>should these been taken from train or validation branch?</p>",
      "rawMarkdown": "does the file [image_ids_and_rotation.csv][1] contain the information about correlation between image names and their id's?\nIs there a sane way to get images sorted to classes in a form of sub-folder structure as for caffe model training?\nI am asking, because what I can see with the\n\n  \n\n     cat image_ids_and_rotation.csv | tail\n\nis the information below from where idk how to determine what is the image id:\n\n&gt; https://farm6.staticflickr.com/3835/15022727040_f6b680d8bc_o.jpg,https://www.flickr.com/photos/davehamster/15022727040,https://creativecommons.org/licenses/by/2.0/,https://www.flickr.com/people/davehamster/,David\n&gt; Merrett,1934 Riley TT\n&gt; Sprite,745972,TYaYy6M4D7qP7X7ZcPc1uw==,https://c2.staticflickr.com/4/3835/15022727040_818e71880c_z.jpg,0\n&gt; 8a23c798ea665d15,train,https://farm6.staticflickr.com/7180/6973157007_6d633ffef3_o.jpg,https://www.flickr.com/photos/mejiaperalta/6973157007,https://creativecommons.org/licenses/by/2.0/,https://www.flickr.com/people/mejiaperalta/,Jorge\n&gt; Mejía\n&gt; \n&gt; peralta,IMG_2936tepe,1214382,qaXSvOxNbkjx6hUOeoYDJg==,https://c4.staticflickr.com/8/7180/6973157007_0cc4687da9_z.jpg,0\n&gt; 90d0031f89de28ee,train,https://c7.staticflickr.com/1/23/27246359_825538fa3e_o.jpg,https://www.flickr.com/photos/emdot/27246359,https://creativecommons.org/licenses/by/2.0/,https://www.flickr.com/people/emdot/,emdot,i\n&gt; swear it was there just a minute\n&gt; ago,561560,i6YWnWUyYlkjTCiOCoQhHA==,https://c6.staticflickr.com/1/23/27246359_825538fa3e_z.jpg?zz=1,0\n\nWhat is the image id in the excerpt above?\n \nupon checking it appears that images id's are aither in the file https://storage.googleapis.com/openimages/2018_04/train/train-annotations-human-imagelabels-boxable.csv or in the file https://storage.googleapis.com/openimages/2018_04/train/train-images-boxable-with-rotation.csv\nThank you.\nImages id's have been located:\n\n    cat train-annotations-human-imagelabels-boxable.csv | tail\nffffeac7d2f37489,verification,/m/07yv9,1\nffffeac7d2f37489,verification,/m/083wq,1\nffffeac7d2f37489,verification,/m/0h9mv,1\nffffeac7d2f37489,verification,/m/0k4j,1\nffffebddbb8bba21,verification,/m/0242l,1\nfffffdaec951185d,verification,/m/01g317,1\nfffffdaec951185d,verification,/m/01xyhv,1\nfffffdaec951185d,verification,/m/03bt1vf,1\nfffffdaec951185d,verification,/m/09j2d,1\nfffffdaec951185d,verification,/m/0dzct,1\n\nshould these been taken from train or validation branch?\n  [1]: https://storage.googleapis.com/openimages/2018_04/image_ids_and_rotation.csv"
    },
    {
      "id": 413174,
      "postDate": "2018-10-31T12:19:00.300Z",
      "rawMarkdown": "",
      "isDeleted": true,
      "replies": [
        {
          "id": 413394,
          "postDate": "2018-10-31T21:06:58.433Z",
          "content": "<p>Hi Amrit,</p>\n\n<p>What you're describing is fine so long as we can reproduce your results from the weights and model that you upload.  You will need to ensure that you've included your stage 2 inference code and detailed instructions on how to run it to generate the predictions that you submit.</p>",
          "rawMarkdown": "Hi Amrit,\n\nWhat you're describing is fine so long as we can reproduce your results from the weights and model that you upload.  You will need to ensure that you've included your stage 2 inference code and detailed instructions on how to run it to generate the predictions that you submit.",
          "votes": 1
        }
      ]
    }
  ],
  "comments": [
    {
      "id": 414888,
      "author_name": "Zineng Tang",
      "author_url": "",
      "post_date": "2018-11-03T20:45:05.330000",
      "content": "<p>Can we upload multiple checkpoints of our model, which basically mean multiple models? Since it is more convenient to make ensembling.</p>\n\n<p>And the code should include stage 1 inference code, and also that for stage 2?</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 412630,
      "author_name": "Sergey Malyutin",
      "author_url": "",
      "post_date": "2018-10-30T14:22:34.373000",
      "content": "<p>Does it mean that we need to upload a script (for example, python script) to generate valid submission file and model weights (for example, *.hdf5 files with weights) and exactly those weights and script must be used to generate predictions for stage 2 ?</p>",
      "votes": 1,
      "replies": [
        {
          "id": 412661,
          "author_name": "Julia Elliott",
          "author_url": "",
          "post_date": "2018-10-30T15:19:18.990000",
          "content": "<p>That's correct.</p>",
          "votes": 2,
          "replies": []
        }
      ]
    },
    {
      "id": 415438,
      "author_name": "Andrei Volodin",
      "author_url": "",
      "post_date": "2018-11-05T05:45:16.287000",
      "content": "<p>does the file <a href=\"https://storage.googleapis.com/openimages/2018_04/image_ids_and_rotation.csv\">image_ids_and_rotation.csv</a> contain the information about correlation between image names and their id's?\nIs there a sane way to get images sorted to classes in a form of sub-folder structure as for caffe model training?\nI am asking, because what I can see with the</p>\n\n<pre><code> cat image_ids_and_rotation.csv | tail\n</code></pre>\n\n<p>is the information below from where idk how to determine what is the image id:</p>\n\n<p>&gt; <a href=\"https://farm6.staticflickr.com/3835/15022727040_f6b680d8bc_o.jpg,https://www.flickr.com/photos/davehamster/15022727040,https://creativecommons.org/licenses/by/2.0/,https://www.flickr.com/people/davehamster/,David\">https://farm6.staticflickr.com/3835/15022727040_f6b680d8bc_o.jpg,https://www.flickr.com/photos/davehamster/15022727040,https://creativecommons.org/licenses/by/2.0/,https://www.flickr.com/people/davehamster/,David</a>\n&gt; Merrett,1934 Riley TT\n&gt; Sprite,745972,TYaYy6M4D7qP7X7ZcPc1uw==,<a href=\"https://c2.staticflickr.com/4/3835/15022727040_818e71880c_z.jpg,0\">https://c2.staticflickr.com/4/3835/15022727040_818e71880c_z.jpg,0</a>\n&gt; 8a23c798ea665d15,train,<a href=\"https://farm6.staticflickr.com/7180/6973157007_6d633ffef3_o.jpg,https://www.flickr.com/photos/mejiaperalta/6973157007,https://creativecommons.org/licenses/by/2.0/,https://www.flickr.com/people/mejiaperalta/,Jorge\">https://farm6.staticflickr.com/7180/6973157007_6d633ffef3_o.jpg,https://www.flickr.com/photos/mejiaperalta/6973157007,https://creativecommons.org/licenses/by/2.0/,https://www.flickr.com/people/mejiaperalta/,Jorge</a>\n&gt; Mejía\n&gt; \n&gt; peralta,IMG_2936tepe,1214382,qaXSvOxNbkjx6hUOeoYDJg==,<a href=\"https://c4.staticflickr.com/8/7180/6973157007_0cc4687da9_z.jpg,0\">https://c4.staticflickr.com/8/7180/6973157007_0cc4687da9_z.jpg,0</a>\n&gt; 90d0031f89de28ee,train,<a href=\"https://c7.staticflickr.com/1/23/27246359_825538fa3e_o.jpg,https://www.flickr.com/photos/emdot/27246359,https://creativecommons.org/licenses/by/2.0/,https://www.flickr.com/people/emdot/,emdot,i\">https://c7.staticflickr.com/1/23/27246359_825538fa3e_o.jpg,https://www.flickr.com/photos/emdot/27246359,https://creativecommons.org/licenses/by/2.0/,https://www.flickr.com/people/emdot/,emdot,i</a>\n&gt; swear it was there just a minute\n&gt; ago,561560,i6YWnWUyYlkjTCiOCoQhHA==,<a href=\"https://c6.staticflickr.com/1/23/27246359_825538fa3e_z.jpg?zz=1,0\">https://c6.staticflickr.com/1/23/27246359_825538fa3e_z.jpg?zz=1,0</a></p>\n\n<p>What is the image id in the excerpt above?</p>\n\n<p>upon checking it appears that images id's are aither in the file <a href=\"https://storage.googleapis.com/openimages/2018_04/train/train-annotations-human-imagelabels-boxable.csv\">https://storage.googleapis.com/openimages/2018_04/train/train-annotations-human-imagelabels-boxable.csv</a> or in the file <a href=\"https://storage.googleapis.com/openimages/2018_04/train/train-images-boxable-with-rotation.csv\">https://storage.googleapis.com/openimages/2018_04/train/train-images-boxable-with-rotation.csv</a>\nThank you.\nImages id's have been located:</p>\n\n<pre><code>cat train-annotations-human-imagelabels-boxable.csv | tail\n</code></pre>\n\n<p>ffffeac7d2f37489,verification,/m/07yv9,1\nffffeac7d2f37489,verification,/m/083wq,1\nffffeac7d2f37489,verification,/m/0h9mv,1\nffffeac7d2f37489,verification,/m/0k4j,1\nffffebddbb8bba21,verification,/m/0242l,1\nfffffdaec951185d,verification,/m/01g317,1\nfffffdaec951185d,verification,/m/01xyhv,1\nfffffdaec951185d,verification,/m/03bt1vf,1\nfffffdaec951185d,verification,/m/09j2d,1\nfffffdaec951185d,verification,/m/0dzct,1</p>\n\n<p>should these been taken from train or validation branch?</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 413174,
      "author_name": "",
      "author_url": "",
      "post_date": "2018-10-31T12:19:00.300000",
      "content": "",
      "votes": 0,
      "replies": [
        {
          "id": 413394,
          "author_name": "James Atwood",
          "author_url": "",
          "post_date": "2018-10-31T21:06:58.433000",
          "content": "<p>Hi Amrit,</p>\n\n<p>What you're describing is fine so long as we can reproduce your results from the weights and model that you upload.  You will need to ensure that you've included your stage 2 inference code and detailed instructions on how to run it to generate the predictions that you submit.</p>",
          "votes": 1,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "412239": "This is your reminder that November 5th (next week) is the close of stage 1 and deadline to upload your model. Model upload can be completed by navigating to the [Team][1] tab, and scrolling down to \"Your Model.\" There will be no exceptions to this deadline, and prize eligibility will be contingent on compliance with this requirement.\n\nUpon the close of stage 1, stage 2 data will be released on November 6th, with only 6 days to submit your predictions on that test set. Note that you will only be permitted to select 1 final submission for scoring. This selected submission is what will be expected to match the uploaded model. See the [timeline][2] page for full details, noting that all deadlines are at 11:59 PM UTC on the corresponding day.\n\nPlease refer to Kaggle's [2-stage competition FAQ][3] for more information about the 2-stage design. However, note the following important distinctions:\n\n - In this 2-stage competition, you are **not permitted** to retrain your models after stage 1. Nor are you permitted to use any data for training, other than [the dataset specified as the eligible train set for this competition][4]. This is an explicit difference from the 2-stage FAQ linked above.\n - It should be noted that the stage 2 test set will contain images that are blurred but do not contain people. This is in line with the [Competition FAQ][5] specifying that competitors should not use dataset peculiarities in their modeling methods.\n - Please ensure compliance with the [Model Eligibility Requirements][6].\n\nBest of luck!\n\n\n  [1]: https://www.kaggle.com/c/inclusive-images-challenge/team\n  [2]: https://www.kaggle.com/c/inclusive-images-challenge#Timeline\n  [3]: https://www.kaggle.com/two-stage-frequently-asked-questions\n  [4]: https://www.kaggle.com/c/inclusive-images-challenge#Data-Download-&amp;-Getting-Started\n  [5]: https://www.kaggle.com/c/inclusive-images-challenge#Inclusive-Images-FAQ\n  [6]: https://www.kaggle.com/c/inclusive-images-challenge#Model-Eligibility-Requirements",
    "414888": "Can we upload multiple checkpoints of our model, which basically mean multiple models? Since it is more convenient to make ensembling.\n\nAnd the code should include stage 1 inference code, and also that for stage 2?",
    "412630": "Does it mean that we need to upload a script (for example, python script) to generate valid submission file and model weights (for example, *.hdf5 files with weights) and exactly those weights and script must be used to generate predictions for stage 2 ?",
    "415438": "does the file [image_ids_and_rotation.csv][1] contain the information about correlation between image names and their id's?\nIs there a sane way to get images sorted to classes in a form of sub-folder structure as for caffe model training?\nI am asking, because what I can see with the\n\n  \n\n     cat image_ids_and_rotation.csv | tail\n\nis the information below from where idk how to determine what is the image id:\n\n&gt; https://farm6.staticflickr.com/3835/15022727040_f6b680d8bc_o.jpg,https://www.flickr.com/photos/davehamster/15022727040,https://creativecommons.org/licenses/by/2.0/,https://www.flickr.com/people/davehamster/,David\n&gt; Merrett,1934 Riley TT\n&gt; Sprite,745972,TYaYy6M4D7qP7X7ZcPc1uw==,https://c2.staticflickr.com/4/3835/15022727040_818e71880c_z.jpg,0\n&gt; 8a23c798ea665d15,train,https://farm6.staticflickr.com/7180/6973157007_6d633ffef3_o.jpg,https://www.flickr.com/photos/mejiaperalta/6973157007,https://creativecommons.org/licenses/by/2.0/,https://www.flickr.com/people/mejiaperalta/,Jorge\n&gt; Mejía\n&gt; \n&gt; peralta,IMG_2936tepe,1214382,qaXSvOxNbkjx6hUOeoYDJg==,https://c4.staticflickr.com/8/7180/6973157007_0cc4687da9_z.jpg,0\n&gt; 90d0031f89de28ee,train,https://c7.staticflickr.com/1/23/27246359_825538fa3e_o.jpg,https://www.flickr.com/photos/emdot/27246359,https://creativecommons.org/licenses/by/2.0/,https://www.flickr.com/people/emdot/,emdot,i\n&gt; swear it was there just a minute\n&gt; ago,561560,i6YWnWUyYlkjTCiOCoQhHA==,https://c6.staticflickr.com/1/23/27246359_825538fa3e_z.jpg?zz=1,0\n\nWhat is the image id in the excerpt above?\n \nupon checking it appears that images id's are aither in the file https://storage.googleapis.com/openimages/2018_04/train/train-annotations-human-imagelabels-boxable.csv or in the file https://storage.googleapis.com/openimages/2018_04/train/train-images-boxable-with-rotation.csv\nThank you.\nImages id's have been located:\n\n    cat train-annotations-human-imagelabels-boxable.csv | tail\nffffeac7d2f37489,verification,/m/07yv9,1\nffffeac7d2f37489,verification,/m/083wq,1\nffffeac7d2f37489,verification,/m/0h9mv,1\nffffeac7d2f37489,verification,/m/0k4j,1\nffffebddbb8bba21,verification,/m/0242l,1\nfffffdaec951185d,verification,/m/01g317,1\nfffffdaec951185d,verification,/m/01xyhv,1\nfffffdaec951185d,verification,/m/03bt1vf,1\nfffffdaec951185d,verification,/m/09j2d,1\nfffffdaec951185d,verification,/m/0dzct,1\n\nshould these been taken from train or validation branch?\n  [1]: https://storage.googleapis.com/openimages/2018_04/image_ids_and_rotation.csv",
    "413174": ""
  }
}