{
  "id": 60411,
  "title": "How to work this many pictures  ?",
  "url": "/competitions/google-ai-open-images-object-detection-track/discussion/60411",
  "author_name": "",
  "post_date": "2018-07-04T06:02:09.559408700Z",
  "votes": null,
  "comment_count": 7,
  "views": 0,
  "content": "<p>Hello , what are the best practices in order to work on this massive amount of data from home ? </p>",
  "messages": [
    {
      "id": "352338",
      "postDate": "07/04/2018 06:02:09",
      "content": "<p>Hello , what are the best practices in order to work on this massive amount of data from home ? </p>",
      "rawMarkdown": "Hello , what are the best practices in order to work on this massive amount of data from home ?",
      "votes": null
    },
    {
      "id": "352356",
      "postDate": "07/04/2018 06:44:10",
      "content": "<p>Use google ML cloud or kaggle, if you want to train on a GPU </p>",
      "rawMarkdown": "Use google ML cloud or kaggle, if you want to train on a GPU",
      "votes": null
    },
    {
      "id": "352383",
      "postDate": "07/04/2018 08:14:39",
      "content": "<p>You could use a datagenerator (e.g. keras.preprocessing.image.flow_from_directory)</p>\n\n<p>Documentation:\n<a href=\"https://keras.io/preprocessing/image/\">https://keras.io/preprocessing/image/</a></p>",
      "rawMarkdown": "You could use a datagenerator (e.g. keras.preprocessing.image.flow_from_directory)\n\nDocumentation:\n[https://keras.io/preprocessing/image/][1]\n\n\n  [1]: https://keras.io/preprocessing/image/",
      "votes": null
    },
    {
      "id": "352601",
      "postDate": "07/04/2018 17:25:40",
      "content": "<p>Is it possible to train so many images on a local GPU? \nif I have a 1080Ti and 1TB HHD.</p>",
      "rawMarkdown": "Is it possible to train so many images on a local GPU? \nif I have a 1080Ti and 1TB HHD.",
      "votes": null
    },
    {
      "id": "352626",
      "postDate": "07/04/2018 18:16:50",
      "content": "<p>Should be ok to build some models, but I am afraid its not enough to get a high ranking</p>",
      "rawMarkdown": "Should be ok to build some models, but I am afraid its not enough to get a high ranking",
      "votes": null
    },
    {
      "id": "352906",
      "postDate": "07/05/2018 11:48:02",
      "content": "<p>Thanks! </p>",
      "rawMarkdown": "Thanks!",
      "votes": null
    },
    {
      "id": "360349",
      "postDate": "07/22/2018 08:02:56",
      "content": "<p>thanks ! </p>",
      "rawMarkdown": "thanks !",
      "votes": null
    },
    {
      "id": "360350",
      "postDate": "07/22/2018 08:03:25",
      "content": "<p>thanks !</p>",
      "rawMarkdown": "thanks !",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 352356,
      "author_name": "saurava57",
      "author_url": "",
      "post_date": "07/04/2018 06:44:10",
      "content": "<p>Use google ML cloud or kaggle, if you want to train on a GPU </p>",
      "votes": null,
      "replies": [
        {
          "id": 360350,
          "author_name": "gregory711",
          "author_url": "",
          "post_date": "07/22/2018 08:03:25",
          "content": "<p>thanks !</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 352383,
      "author_name": "christofhenkel",
      "author_url": "",
      "post_date": "07/04/2018 08:14:39",
      "content": "<p>You could use a datagenerator (e.g. keras.preprocessing.image.flow_from_directory)</p>\n\n<p>Documentation:\n<a href=\"https://keras.io/preprocessing/image/\">https://keras.io/preprocessing/image/</a></p>",
      "votes": null,
      "replies": [
        {
          "id": 360349,
          "author_name": "gregory711",
          "author_url": "",
          "post_date": "07/22/2018 08:02:56",
          "content": "<p>thanks ! </p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 352601,
      "author_name": "yezizhu",
      "author_url": "",
      "post_date": "07/04/2018 17:25:40",
      "content": "<p>Is it possible to train so many images on a local GPU? \nif I have a 1080Ti and 1TB HHD.</p>",
      "votes": null,
      "replies": [
        {
          "id": 352626,
          "author_name": "christofhenkel",
          "author_url": "",
          "post_date": "07/04/2018 18:16:50",
          "content": "<p>Should be ok to build some models, but I am afraid its not enough to get a high ranking</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 352906,
          "author_name": "yezizhu",
          "author_url": "",
          "post_date": "07/05/2018 11:48:02",
          "content": "<p>Thanks! </p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "352338": "Hello , what are the best practices in order to work on this massive amount of data from home ?",
    "352356": "Use google ML cloud or kaggle, if you want to train on a GPU",
    "352383": "You could use a datagenerator (e.g. keras.preprocessing.image.flow_from_directory)\n\nDocumentation:\n[https://keras.io/preprocessing/image/][1]\n\n\n  [1]: https://keras.io/preprocessing/image/",
    "352601": "Is it possible to train so many images on a local GPU? \nif I have a 1080Ti and 1TB HHD.",
    "352626": "Should be ok to build some models, but I am afraid its not enough to get a high ranking",
    "352906": "Thanks!",
    "360349": "thanks !",
    "360350": "thanks !"
  },
  "source": "meta"
}