{
  "id": 68668,
  "title": "Best model performance on non-empty images",
  "url": "/competitions/airbus-ship-detection/discussion/68668",
  "author_name": "",
  "post_date": "2018-10-15T18:53:24.412319300Z",
  "votes": 3,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Hi everyone.</p>\n\n<p>I am training a semantic segmentation model on non-empty images and I am currently at <code>CV 0.3</code>. </p>\n\n<p><a href=\"https://app.neptune.ml/-/dashboard/experiment/14318634-6228-4357-b300-2337b38f0e25\"><img src=\"https://gist.githubusercontent.com/jakubczakon/cac72983726a970690ba7c33708e100b/raw/6939d56fab981bebb8d89c8cd4f37f6da4b431e3/lkm_densenet.png\" alt=\"image\"></a></p>\n\n<p>The <a href=\"https://app.neptune.ml/-/dashboard/experiment/14318634-6228-4357-b300-2337b38f0e25\">experiment</a> is still running but I'd figured I'd ask already.</p>\n\n<p>What is the performance (<code>f2 score</code>) of your best models on non-empty images?</p>",
  "messages": [
    {
      "id": "404472",
      "postDate": "10/15/2018 18:53:24",
      "content": "<p>Hi everyone.</p>\n\n<p>I am training a semantic segmentation model on non-empty images and I am currently at <code>CV 0.3</code>. </p>\n\n<p><a href=\"https://app.neptune.ml/-/dashboard/experiment/14318634-6228-4357-b300-2337b38f0e25\"><img src=\"https://gist.githubusercontent.com/jakubczakon/cac72983726a970690ba7c33708e100b/raw/6939d56fab981bebb8d89c8cd4f37f6da4b431e3/lkm_densenet.png\" alt=\"image\"></a></p>\n\n<p>The <a href=\"https://app.neptune.ml/-/dashboard/experiment/14318634-6228-4357-b300-2337b38f0e25\">experiment</a> is still running but I'd figured I'd ask already.</p>\n\n<p>What is the performance (<code>f2 score</code>) of your best models on non-empty images?</p>",
      "rawMarkdown": "Hi everyone.\n\nI am training a semantic segmentation model on non-empty images and I am currently at `CV 0.3`. \n\n[![image](https://gist.githubusercontent.com/jakubczakon/cac72983726a970690ba7c33708e100b/raw/6939d56fab981bebb8d89c8cd4f37f6da4b431e3/lkm_densenet.png)](https://app.neptune.ml/-/dashboard/experiment/14318634-6228-4357-b300-2337b38f0e25)\n\nThe [experiment](https://app.neptune.ml/-/dashboard/experiment/14318634-6228-4357-b300-2337b38f0e25) is still running but I'd figured I'd ask already.\n\nWhat is the performance (`f2 score`) of your best models on non-empty images?",
      "votes": null
    },
    {
      "id": "411342",
      "postDate": "10/27/2018 22:25:40",
      "content": "<p>0.8321 on single fold (all fold is with non-empty images)</p>",
      "rawMarkdown": "0.8321 on single fold (all fold is with non-empty images)",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 411342,
      "author_name": "maksimovka",
      "author_url": "",
      "post_date": "10/27/2018 22:25:40",
      "content": "<p>0.8321 on single fold (all fold is with non-empty images)</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "404472": "Hi everyone.\n\nI am training a semantic segmentation model on non-empty images and I am currently at `CV 0.3`. \n\n[![image](https://gist.githubusercontent.com/jakubczakon/cac72983726a970690ba7c33708e100b/raw/6939d56fab981bebb8d89c8cd4f37f6da4b431e3/lkm_densenet.png)](https://app.neptune.ml/-/dashboard/experiment/14318634-6228-4357-b300-2337b38f0e25)\n\nThe [experiment](https://app.neptune.ml/-/dashboard/experiment/14318634-6228-4357-b300-2337b38f0e25) is still running but I'd figured I'd ask already.\n\nWhat is the performance (`f2 score`) of your best models on non-empty images?",
    "411342": "0.8321 on single fold (all fold is with non-empty images)"
  },
  "source": "meta"
}