{
  "id": 54149,
  "title": "Dice Score vs IOU",
  "url": "/competitions/data-science-bowl-2018/discussion/54149",
  "author_name": "",
  "post_date": "2018-04-10T08:31:15.750535600Z",
  "votes": 2,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Hi Guys.  I know its late but Im still trying to get a network learning hopefully by the deadline to learn how to work with these kind of problems.  I am confused between Dice Score and IOU.  It seems Dice Score just adds a TP to the numerator and denominator. IOU doesn't take this into account.  Do we just use one or the other for our loss function?\nAlso, calculating the IOU gives an overall value for the whole image, but individual gradient values to update the image pixels in each epoch?  Thanks!</p>",
  "messages": [
    {
      "id": "311549",
      "postDate": "04/10/2018 08:31:15",
      "content": "<p>Hi Guys.  I know its late but Im still trying to get a network learning hopefully by the deadline to learn how to work with these kind of problems.  I am confused between Dice Score and IOU.  It seems Dice Score just adds a TP to the numerator and denominator. IOU doesn't take this into account.  Do we just use one or the other for our loss function?\nAlso, calculating the IOU gives an overall value for the whole image, but individual gradient values to update the image pixels in each epoch?  Thanks!</p>",
      "rawMarkdown": "Hi Guys.  I know its late but Im still trying to get a network learning hopefully by the deadline to learn how to work with these kind of problems.  I am confused between Dice Score and IOU.  It seems Dice Score just adds a TP to the numerator and denominator. IOU doesn't take this into account.  Do we just use one or the other for our loss function?\nAlso, calculating the IOU gives an overall value for the whole image, but individual gradient values to update the image pixels in each epoch?  Thanks!",
      "votes": null
    },
    {
      "id": "624774",
      "postDate": "09/12/2019 11:00:34",
      "content": "<p>Check this '<a href=\"https://stackoverflow.com/questions/40475246/why-does-one-not-use-iou-for-training\">Why does one not use IOU for training?</a>' and also this <a href=\"https://stats.stackexchange.com/questions/273537/f1-dice-score-vs-iou\">'F1/Dice-Score vs IoU'</a></p>",
      "rawMarkdown": "Check this '[Why does one not use IOU for training?](https://stackoverflow.com/questions/40475246/why-does-one-not-use-iou-for-training)' and also this ['F1/Dice-Score vs IoU'](https://stats.stackexchange.com/questions/273537/f1-dice-score-vs-iou)",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 624774,
      "author_name": "ipythonx",
      "author_url": "",
      "post_date": "09/12/2019 11:00:34",
      "content": "<p>Check this '<a href=\"https://stackoverflow.com/questions/40475246/why-does-one-not-use-iou-for-training\">Why does one not use IOU for training?</a>' and also this <a href=\"https://stats.stackexchange.com/questions/273537/f1-dice-score-vs-iou\">'F1/Dice-Score vs IoU'</a></p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "311549": "Hi Guys.  I know its late but Im still trying to get a network learning hopefully by the deadline to learn how to work with these kind of problems.  I am confused between Dice Score and IOU.  It seems Dice Score just adds a TP to the numerator and denominator. IOU doesn't take this into account.  Do we just use one or the other for our loss function?\nAlso, calculating the IOU gives an overall value for the whole image, but individual gradient values to update the image pixels in each epoch?  Thanks!",
    "624774": "Check this '[Why does one not use IOU for training?](https://stackoverflow.com/questions/40475246/why-does-one-not-use-iou-for-training)' and also this ['F1/Dice-Score vs IoU'](https://stats.stackexchange.com/questions/273537/f1-dice-score-vs-iou)"
  },
  "source": "meta"
}