{
  "id": 156489,
  "title": "[WBF over TTA][Single Model] EfficientDet",
  "url": "/competitions/global-wheat-detection/discussion/156489",
  "author_name": "",
  "post_date": "2020-06-06T11:25:26.511775800Z",
  "votes": 17,
  "comment_count": 10,
  "views": 0,
  "content": "<p>Hi everyone!</p>\n\n<p>I noticed you were bored without my kernels :D</p>\n\n<p>So, today I would like to share with you, my friends, kernel about TTA with using WBF:</p>\n\n<ul>\n<li><a href=\"https://www.kaggle.com/shonenkov/wbf-over-tta-single-model-efficientdet\">[WBF over TTA][Single Model] EfficientDet</a></li>\n</ul>\n\n<p>I used single public model from <a href=\"https://www.kaggle.com/shonenkov/inference-efficientdet\">this kernel about EfficientDet</a>.</p>\n\n<p>I created simple TTA API for clear understanding how it works!</p>\n\n<p>I have got some improvements:\n<code>\n0.7193 --&gt; 0.7294 (+1%)\n</code>\nWelcome!</p>",
  "messages": [
    {
      "id": "876036",
      "postDate": "06/06/2020 11:25:26",
      "content": "<p>Hi everyone!</p>\n\n<p>I noticed you were bored without my kernels :D</p>\n\n<p>So, today I would like to share with you, my friends, kernel about TTA with using WBF:</p>\n\n<ul>\n<li><a href=\"https://www.kaggle.com/shonenkov/wbf-over-tta-single-model-efficientdet\">[WBF over TTA][Single Model] EfficientDet</a></li>\n</ul>\n\n<p>I used single public model from <a href=\"https://www.kaggle.com/shonenkov/inference-efficientdet\">this kernel about EfficientDet</a>.</p>\n\n<p>I created simple TTA API for clear understanding how it works!</p>\n\n<p>I have got some improvements:\n<code>\n0.7193 --&gt; 0.7294 (+1%)\n</code>\nWelcome!</p>",
      "rawMarkdown": "Hi everyone!\n\nI noticed you were bored without my kernels :D\n\nSo, today I would like to share with you, my friends, kernel about TTA with using WBF:\n\n- [[WBF over TTA][Single Model] EfficientDet](https://www.kaggle.com/shonenkov/wbf-over-tta-single-model-efficientdet)\n\nI used single public model from [this kernel about EfficientDet](https://www.kaggle.com/shonenkov/inference-efficientdet).\n\nI created simple TTA API for clear understanding how it works!\n\n\nI have got some improvements:\n```\n0.7193 --&gt; 0.7294 (+1%)\n```\nWelcome!",
      "votes": null
    },
    {
      "id": "876156",
      "postDate": "06/06/2020 13:43:36",
      "content": "<p>I've been using this trick! Definitely gave me a boost</p>",
      "rawMarkdown": "I've been using this trick! Definitely gave me a boost",
      "votes": null
    },
    {
      "id": "876259",
      "postDate": "06/06/2020 15:21:43",
      "content": "<p>thank you!</p>",
      "rawMarkdown": "thank you!",
      "votes": null
    },
    {
      "id": "876289",
      "postDate": "06/06/2020 15:49:33",
      "content": "<p>Great job! Not as much as you, but it also helps me to increase 0.005 :))</p>",
      "rawMarkdown": "Great job! Not as much as you, but it also helps me to increase 0.005 :))",
      "votes": null
    },
    {
      "id": "876335",
      "postDate": "06/06/2020 16:19:12",
      "content": "<p>welcome ;)</p>",
      "rawMarkdown": "welcome ;)",
      "votes": null
    },
    {
      "id": "876346",
      "postDate": "06/06/2020 16:27:22",
      "content": "<p>Hi <a href=\"/shonenkov\">@shonenkov</a>, you have tried TTA for your Bayesian optimization WBF method?</p>",
      "rawMarkdown": "Hi @shonenkov, you have tried TTA for your Bayesian optimization WBF method?",
      "votes": null
    },
    {
      "id": "876381",
      "postDate": "06/06/2020 16:43:59",
      "content": "<p><a href=\"/duykhanh99\">@duykhanh99</a> not yet, only for single model :)</p>",
      "rawMarkdown": "duykhanh99 not yet, only for single model :)",
      "votes": null
    },
    {
      "id": "876405",
      "postDate": "06/06/2020 16:57:38",
      "content": "<p>Thanks!</p>",
      "rawMarkdown": "Thanks!",
      "votes": null
    },
    {
      "id": "877544",
      "postDate": "06/07/2020 17:25:42",
      "content": "<p>Thanks for your share again!</p>",
      "rawMarkdown": "Thanks for your share again!",
      "votes": null
    },
    {
      "id": "877769",
      "postDate": "06/08/2020 00:35:29",
      "content": "<p>Thanks for sharing!!!</p>",
      "rawMarkdown": "Thanks for sharing!!!",
      "votes": null
    },
    {
      "id": "879244",
      "postDate": "06/09/2020 10:34:46",
      "content": "<p>Great job! For my single model, I have got 0.6902-&gt;0.7126.\nThanks for sharing!</p>",
      "rawMarkdown": "Great job! For my single model, I have got 0.6902-&gt;0.7126.\nThanks for sharing!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 876156,
      "author_name": "matthewmasters",
      "author_url": "",
      "post_date": "06/06/2020 13:43:36",
      "content": "<p>I've been using this trick! Definitely gave me a boost</p>",
      "votes": null,
      "replies": [
        {
          "id": 876259,
          "author_name": "shonenkov",
          "author_url": "",
          "post_date": "06/06/2020 15:21:43",
          "content": "<p>thank you!</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 876289,
      "author_name": "duykhanh99",
      "author_url": "",
      "post_date": "06/06/2020 15:49:33",
      "content": "<p>Great job! Not as much as you, but it also helps me to increase 0.005 :))</p>",
      "votes": null,
      "replies": [
        {
          "id": 876335,
          "author_name": "shonenkov",
          "author_url": "",
          "post_date": "06/06/2020 16:19:12",
          "content": "<p>welcome ;)</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 876346,
          "author_name": "duykhanh99",
          "author_url": "",
          "post_date": "06/06/2020 16:27:22",
          "content": "<p>Hi <a href=\"/shonenkov\">@shonenkov</a>, you have tried TTA for your Bayesian optimization WBF method?</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 876381,
          "author_name": "shonenkov",
          "author_url": "",
          "post_date": "06/06/2020 16:43:59",
          "content": "<p><a href=\"/duykhanh99\">@duykhanh99</a> not yet, only for single model :)</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 876405,
          "author_name": "duykhanh99",
          "author_url": "",
          "post_date": "06/06/2020 16:57:38",
          "content": "<p>Thanks!</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 877544,
      "author_name": "adityaecdrid",
      "author_url": "",
      "post_date": "06/07/2020 17:25:42",
      "content": "<p>Thanks for your share again!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 877769,
      "author_name": "spidyweb",
      "author_url": "",
      "post_date": "06/08/2020 00:35:29",
      "content": "<p>Thanks for sharing!!!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 879244,
      "author_name": "ufownl",
      "author_url": "",
      "post_date": "06/09/2020 10:34:46",
      "content": "<p>Great job! For my single model, I have got 0.6902-&gt;0.7126.\nThanks for sharing!</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "876036": "Hi everyone!\n\nI noticed you were bored without my kernels :D\n\nSo, today I would like to share with you, my friends, kernel about TTA with using WBF:\n\n- [[WBF over TTA][Single Model] EfficientDet](https://www.kaggle.com/shonenkov/wbf-over-tta-single-model-efficientdet)\n\nI used single public model from [this kernel about EfficientDet](https://www.kaggle.com/shonenkov/inference-efficientdet).\n\nI created simple TTA API for clear understanding how it works!\n\n\nI have got some improvements:\n```\n0.7193 --&gt; 0.7294 (+1%)\n```\nWelcome!",
    "876156": "I've been using this trick! Definitely gave me a boost",
    "876259": "thank you!",
    "876289": "Great job! Not as much as you, but it also helps me to increase 0.005 :))",
    "876335": "welcome ;)",
    "876346": "Hi @shonenkov, you have tried TTA for your Bayesian optimization WBF method?",
    "876381": "duykhanh99 not yet, only for single model :)",
    "876405": "Thanks!",
    "877544": "Thanks for your share again!",
    "877769": "Thanks for sharing!!!",
    "879244": "Great job! For my single model, I have got 0.6902-&gt;0.7126.\nThanks for sharing!"
  },
  "source": "meta"
}