{
  "id": 248015,
  "title": "Best image level score",
  "url": "/competitions/siim-covid19-detection/discussion/248015",
  "author_name": "",
  "post_date": "2021-06-22T04:53:19.349295300Z",
  "votes": 3,
  "comment_count": 5,
  "views": 0,
  "content": "<p>Hey guys i wanted to know what is best possible score by only submitting image level prediction( all study level as negative) I have used Yolov5 small to attain the score 0.23</p>",
  "messages": [
    {
      "id": "1360421",
      "postDate": "06/22/2021 04:53:19",
      "content": "<p>Hey guys i wanted to know what is best possible score by only submitting image level prediction( all study level as negative) I have used Yolov5 small to attain the score 0.23</p>",
      "rawMarkdown": "Hey guys i wanted to know what is best possible score by only submitting image level prediction( all study level as negative) I have used Yolov5 small to attain the score 0.23",
      "votes": null
    },
    {
      "id": "1361720",
      "postDate": "06/23/2021 03:18:16",
      "content": "<p>I'm not sure about the image score. <br>\nBut you can fill study level with null string to get the true score of image level.<br>\nIt will be easier to compare .</p>",
      "rawMarkdown": "I'm not sure about the image score. \nBut you can fill study level with null string to get the true score of image level.\nIt will be easier to compare .",
      "votes": null
    },
    {
      "id": "1362685",
      "postDate": "06/23/2021 15:10:45",
      "content": "<p>Hey, According to my calculations the your actual score will be (0.18 + Study Level score) As the sample submission before rescoring scored 0.05 so that is pretty much all there was to study level without change.</p>\n<p>Basically if my calculations are correct your score is good…</p>",
      "rawMarkdown": "Hey, According to my calculations the your actual score will be (0.18 + Study Level score) As the sample submission before rescoring scored 0.05 so that is pretty much all there was to study level without change.\n\nBasically if my calculations are correct your score is good...",
      "votes": null
    },
    {
      "id": "1362701",
      "postDate": "06/23/2021 15:18:07",
      "content": "<p>When I calculate my mAP I only get 0.20…..I dont know whether there is something wrong..</p>",
      "rawMarkdown": "When I calculate my mAP I only get 0.20.....I dont know whether there is something wrong..",
      "votes": null
    },
    {
      "id": "1362807",
      "postDate": "06/23/2021 16:51:38",
      "content": "<p>You can take my code to calculate it, so basically mAP for study level.</p>\n<p>from sklearn.metrics import average_precision_score<br>\nscore = 0<br>\nfor x in range(4): score += average_precision_score(tst_lbls[:, x], tst_outs[:, x]) / 4 * 2/3</p>\n<p>tst_lbls is labels in shape (N, 4) and tst_outs are model outputs in shape of (N, 4), N being the number of samples.</p>",
      "rawMarkdown": "You can take my code to calculate it, so basically mAP for study level.\n\nfrom sklearn.metrics import average_precision_score\nscore = 0\nfor x in range(4): score += average_precision_score(tst_lbls[:, x], tst_outs[:, x]) / 4 * 2/3\n\ntst_lbls is labels in shape (N, 4) and tst_outs are model outputs in shape of (N, 4), N being the number of samples.",
      "votes": null
    },
    {
      "id": "1363152",
      "postDate": "06/24/2021 01:07:02",
      "content": "<p>Thanks.But it is only for study level.</p>",
      "rawMarkdown": "Thanks.But it is only for study level.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1361720,
      "author_name": "menghsuanliu",
      "author_url": "",
      "post_date": "06/23/2021 03:18:16",
      "content": "<p>I'm not sure about the image score. <br>\nBut you can fill study level with null string to get the true score of image level.<br>\nIt will be easier to compare .</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1362685,
      "author_name": "harshitsheoran",
      "author_url": "",
      "post_date": "06/23/2021 15:10:45",
      "content": "<p>Hey, According to my calculations the your actual score will be (0.18 + Study Level score) As the sample submission before rescoring scored 0.05 so that is pretty much all there was to study level without change.</p>\n<p>Basically if my calculations are correct your score is good…</p>",
      "votes": null,
      "replies": [
        {
          "id": 1362701,
          "author_name": "zekunn",
          "author_url": "",
          "post_date": "06/23/2021 15:18:07",
          "content": "<p>When I calculate my mAP I only get 0.20…..I dont know whether there is something wrong..</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1362807,
          "author_name": "harshitsheoran",
          "author_url": "",
          "post_date": "06/23/2021 16:51:38",
          "content": "<p>You can take my code to calculate it, so basically mAP for study level.</p>\n<p>from sklearn.metrics import average_precision_score<br>\nscore = 0<br>\nfor x in range(4): score += average_precision_score(tst_lbls[:, x], tst_outs[:, x]) / 4 * 2/3</p>\n<p>tst_lbls is labels in shape (N, 4) and tst_outs are model outputs in shape of (N, 4), N being the number of samples.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1363152,
          "author_name": "zekunn",
          "author_url": "",
          "post_date": "06/24/2021 01:07:02",
          "content": "<p>Thanks.But it is only for study level.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1360421": "Hey guys i wanted to know what is best possible score by only submitting image level prediction( all study level as negative) I have used Yolov5 small to attain the score 0.23",
    "1361720": "I'm not sure about the image score. \nBut you can fill study level with null string to get the true score of image level.\nIt will be easier to compare .",
    "1362685": "Hey, According to my calculations the your actual score will be (0.18 + Study Level score) As the sample submission before rescoring scored 0.05 so that is pretty much all there was to study level without change.\n\nBasically if my calculations are correct your score is good...",
    "1362701": "When I calculate my mAP I only get 0.20.....I dont know whether there is something wrong..",
    "1362807": "You can take my code to calculate it, so basically mAP for study level.\n\nfrom sklearn.metrics import average_precision_score\nscore = 0\nfor x in range(4): score += average_precision_score(tst_lbls[:, x], tst_outs[:, x]) / 4 * 2/3\n\ntst_lbls is labels in shape (N, 4) and tst_outs are model outputs in shape of (N, 4), N being the number of samples.",
    "1363152": "Thanks.But it is only for study level."
  },
  "source": "meta"
}