{
  "id": 242843,
  "title": "2nd place solution",
  "url": "/competitions/plant-pathology-2021-fgvc8/discussion/242843",
  "author_name": "Jianye He",
  "post_date": "2021-05-31T06:33:10.761000",
  "votes": 17,
  "comment_count": 2,
  "views": 0,
  "content": "<p><strong>1. Data</strong> FGVC7+FGVC8, 224x224/384x384<br>\n<strong>2.model</strong> SEResNeXt50 + SwinT<br>\n<strong>3.Training</strong>  BCE loss, Soften the label, hflip, shiftscalerotate, colorjitter<br>\n<strong>4.Inference</strong> ensemble (serx50+swint), no tta, no weight, no cv(these tricks just don't work in my exps);Best single model is from serx50(public:0.86718, private:0.87643)</p>",
  "messages": [
    {
      "id": 1329442,
      "postDate": "2021-05-31T06:33:10.760Z",
      "content": "<p><strong>1. Data</strong> FGVC7+FGVC8, 224x224/384x384<br>\n<strong>2.model</strong> SEResNeXt50 + SwinT<br>\n<strong>3.Training</strong>  BCE loss, Soften the label, hflip, shiftscalerotate, colorjitter<br>\n<strong>4.Inference</strong> ensemble (serx50+swint), no tta, no weight, no cv(these tricks just don't work in my exps);Best single model is from serx50(public:0.86718, private:0.87643)</p>",
      "rawMarkdown": "**1. Data** FGVC7+FGVC8, 224x224/384x384\n**2.model** SEResNeXt50 + SwinT\n**3.Training**  BCE loss, Soften the label, hflip, shiftscalerotate, colorjitter\n**4.Inference** ensemble (serx50+swint), no tta, no weight, no cv(these tricks just don't work in my exps);Best single model is from serx50(public:0.86718, private:0.87643)",
      "votes": 17
    },
    {
      "id": 1368361,
      "postDate": "2021-06-28T14:09:58.500Z",
      "content": "<p>Hi, <br>\nThanks for participation and congratulations for the 2nd place! Could you please send us email/contact us to organize a brief meeting to learn about your code and solution. Thanks. </p>\n<p>Plant Pathology 2021 Host</p>",
      "rawMarkdown": "Hi, \nThanks for participation and congratulations for the 2nd place! Could you please send us email/contact us to organize a brief meeting to learn about your code and solution. Thanks. \n\nPlant Pathology 2021 Host"
    },
    {
      "id": 1369233,
      "postDate": "2021-06-29T08:20:34.440Z",
      "rawMarkdown": "",
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 1368361,
      "author_name": "Fruit Pathology",
      "author_url": "",
      "post_date": "2021-06-28T14:09:58.500000",
      "content": "<p>Hi, <br>\nThanks for participation and congratulations for the 2nd place! Could you please send us email/contact us to organize a brief meeting to learn about your code and solution. Thanks. </p>\n<p>Plant Pathology 2021 Host</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 1369233,
      "author_name": "",
      "author_url": "",
      "post_date": "2021-06-29T08:20:34.440000",
      "content": "",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1329442": "**1. Data** FGVC7+FGVC8, 224x224/384x384\n**2.model** SEResNeXt50 + SwinT\n**3.Training**  BCE loss, Soften the label, hflip, shiftscalerotate, colorjitter\n**4.Inference** ensemble (serx50+swint), no tta, no weight, no cv(these tricks just don't work in my exps);Best single model is from serx50(public:0.86718, private:0.87643)",
    "1368361": "Hi, \nThanks for participation and congratulations for the 2nd place! Could you please send us email/contact us to organize a brief meeting to learn about your code and solution. Thanks. \n\nPlant Pathology 2021 Host",
    "1369233": ""
  }
}