{
  "id": 34091,
  "title": "Did anybody have a baseline performance?",
  "url": "/competitions/inaturalist-challenge-at-fgvc-2017/discussion/34091",
  "author_name": "Joy",
  "post_date": "2017-06-03T11:28:49.719000",
  "votes": 0,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Did anybody have a baseline performance?</p>",
  "messages": [
    {
      "id": 188685,
      "postDate": "2017-06-03T17:38:15.203Z",
      "content": "<p>@iFighting Here is a write up that describes an Inception V3 model that can get about 85% top five accuracy on the validation set. There is also links to a pretrained model included: <a href=\"https://medium.com/@macaodha/inaturalist-2017-species-classification-challenge-4ff4d1499279\">https://medium.com/@macaodha/inaturalist-2017-species-classification-challenge-4ff4d1499279</a></p>",
      "rawMarkdown": "@iFighting Here is a write up that describes an Inception V3 model that can get about 85% top five accuracy on the validation set. There is also links to a pretrained model included: https://medium.com/@macaodha/inaturalist-2017-species-classification-challenge-4ff4d1499279",
      "replies": [
        {
          "id": 188757,
          "postDate": "2017-06-04T01:08:35.393Z",
          "content": "<p>i have try several methods(including inception v4, v3, resnet),i only can get about 80% @top5 acc.</p>",
          "rawMarkdown": "i have try several methods(including inception v4, v3, resnet),i only can get about 80% @top5 acc."
        }
      ]
    },
    {
      "id": 188613,
      "postDate": "2017-06-03T11:28:49.720Z",
      "content": "<p>Did anybody have a baseline performance?</p>",
      "rawMarkdown": "Did anybody have a baseline performance?\n"
    }
  ],
  "comments": [
    {
      "id": 188685,
      "author_name": "macaodha",
      "author_url": "",
      "post_date": "2017-06-03T17:38:15.203000",
      "content": "<p>@iFighting Here is a write up that describes an Inception V3 model that can get about 85% top five accuracy on the validation set. There is also links to a pretrained model included: <a href=\"https://medium.com/@macaodha/inaturalist-2017-species-classification-challenge-4ff4d1499279\">https://medium.com/@macaodha/inaturalist-2017-species-classification-challenge-4ff4d1499279</a></p>",
      "votes": 0,
      "replies": [
        {
          "id": 188757,
          "author_name": "Joy",
          "author_url": "",
          "post_date": "2017-06-04T01:08:35.393000",
          "content": "<p>i have try several methods(including inception v4, v3, resnet),i only can get about 80% @top5 acc.</p>",
          "votes": 0,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "188685": "@iFighting Here is a write up that describes an Inception V3 model that can get about 85% top five accuracy on the validation set. There is also links to a pretrained model included: https://medium.com/@macaodha/inaturalist-2017-species-classification-challenge-4ff4d1499279",
    "188613": "Did anybody have a baseline performance?\n"
  }
}