{
  "id": 98453,
  "title": "Be careful about your augmentation...",
  "url": "/competitions/aptos2019-blindness-detection/discussion/98453",
  "author_name": "",
  "post_date": "2019-07-03T21:36:42.324940800Z",
  "votes": 2,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Probably a useless reminder but just in case: \ndo pay attention to the way you do augmentation, or say, manipulate your image data.</p>\n\n<p>Each library has different default behaviors in terms of image operations like read (opencv, PIL), save, round and so on. Popular augmentation libs like albumentation / imgaug also have a lot of different behaviors. </p>\n\n<p>I have a framework runs perfectly fine with my local env but I have to ship a bunch of code pieces to the kernel due to the kernel competition setting then I encountered several problems...</p>",
  "messages": [
    {
      "id": "567705",
      "postDate": "07/03/2019 21:36:42",
      "content": "<p>Probably a useless reminder but just in case: \ndo pay attention to the way you do augmentation, or say, manipulate your image data.</p>\n\n<p>Each library has different default behaviors in terms of image operations like read (opencv, PIL), save, round and so on. Popular augmentation libs like albumentation / imgaug also have a lot of different behaviors. </p>\n\n<p>I have a framework runs perfectly fine with my local env but I have to ship a bunch of code pieces to the kernel due to the kernel competition setting then I encountered several problems...</p>",
      "rawMarkdown": "Probably a useless reminder but just in case: \ndo pay attention to the way you do augmentation, or say, manipulate your image data.\n\nEach library has different default behaviors in terms of image operations like read (opencv, PIL), save, round and so on. Popular augmentation libs like albumentation / imgaug also have a lot of different behaviors. \n\nI have a framework runs perfectly fine with my local env but I have to ship a bunch of code pieces to the kernel due to the kernel competition setting then I encountered several problems...",
      "votes": null
    },
    {
      "id": "567869",
      "postDate": "07/04/2019 05:10:29",
      "content": "<p>thanks your advice!</p>",
      "rawMarkdown": "thanks your advice!",
      "votes": null
    },
    {
      "id": "570870",
      "postDate": "07/08/2019 21:27:19",
      "content": "<p>Thanks for your attention. Have a good day.</p>",
      "rawMarkdown": "Thanks for your attention. Have a good day.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 567869,
      "author_name": "leixiang",
      "author_url": "",
      "post_date": "07/04/2019 05:10:29",
      "content": "<p>thanks your advice!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 570870,
      "author_name": "burakarslan38",
      "author_url": "",
      "post_date": "07/08/2019 21:27:19",
      "content": "<p>Thanks for your attention. Have a good day.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "567705": "Probably a useless reminder but just in case: \ndo pay attention to the way you do augmentation, or say, manipulate your image data.\n\nEach library has different default behaviors in terms of image operations like read (opencv, PIL), save, round and so on. Popular augmentation libs like albumentation / imgaug also have a lot of different behaviors. \n\nI have a framework runs perfectly fine with my local env but I have to ship a bunch of code pieces to the kernel due to the kernel competition setting then I encountered several problems...",
    "567869": "thanks your advice!",
    "570870": "Thanks for your attention. Have a good day."
  },
  "source": "meta"
}