{
  "id": 125463,
  "title": "Compensating for head motion",
  "url": "/competitions/deepfake-detection-challenge/discussion/125463",
  "author_name": "",
  "post_date": "2020-01-10T20:03:12.897985300Z",
  "votes": null,
  "comment_count": 2,
  "views": 0,
  "content": "<p>I'm using mtcnn and I am happy with the results. However, one extra thing I'd like to do is to adjust the boxes to compensate as much as possible for bulk head motion as opposed to motion of facial features such as eyes, mouth, etc. Ant suggestions for a way to do this, with any package? My naive take would be correlation but anything  I write will likely be expensive.</p>",
  "messages": [
    {
      "id": "715776",
      "postDate": "01/10/2020 20:03:12",
      "content": "<p>I'm using mtcnn and I am happy with the results. However, one extra thing I'd like to do is to adjust the boxes to compensate as much as possible for bulk head motion as opposed to motion of facial features such as eyes, mouth, etc. Ant suggestions for a way to do this, with any package? My naive take would be correlation but anything  I write will likely be expensive.</p>",
      "rawMarkdown": "I'm using mtcnn and I am happy with the results. However, one extra thing I'd like to do is to adjust the boxes to compensate as much as possible for bulk head motion as opposed to motion of facial features such as eyes, mouth, etc. Ant suggestions for a way to do this, with any package? My naive take would be correlation but anything  I write will likely be expensive.",
      "votes": null
    },
    {
      "id": "715914",
      "postDate": "01/11/2020 00:56:36",
      "content": "<p>Look at dlib's object tracking. Give it a bounding box and it will try and track it using correlation.</p>\n\n<p><a href=\"https://www.pyimagesearch.com/2018/10/22/object-tracking-with-dlib/\">https://www.pyimagesearch.com/2018/10/22/object-tracking-with-dlib/</a></p>",
      "rawMarkdown": "Look at dlib's object tracking. Give it a bounding box and it will try and track it using correlation.\n\nhttps://www.pyimagesearch.com/2018/10/22/object-tracking-with-dlib/",
      "votes": null
    },
    {
      "id": "715936",
      "postDate": "01/11/2020 02:09:01",
      "content": "<p>Opencv has very good tracker and multitracker</p>",
      "rawMarkdown": "Opencv has very good tracker and multitracker",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 715914,
      "author_name": "maralski",
      "author_url": "",
      "post_date": "01/11/2020 00:56:36",
      "content": "<p>Look at dlib's object tracking. Give it a bounding box and it will try and track it using correlation.</p>\n\n<p><a href=\"https://www.pyimagesearch.com/2018/10/22/object-tracking-with-dlib/\">https://www.pyimagesearch.com/2018/10/22/object-tracking-with-dlib/</a></p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 715936,
      "author_name": "moshel",
      "author_url": "",
      "post_date": "01/11/2020 02:09:01",
      "content": "<p>Opencv has very good tracker and multitracker</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "715776": "I'm using mtcnn and I am happy with the results. However, one extra thing I'd like to do is to adjust the boxes to compensate as much as possible for bulk head motion as opposed to motion of facial features such as eyes, mouth, etc. Ant suggestions for a way to do this, with any package? My naive take would be correlation but anything  I write will likely be expensive.",
    "715914": "Look at dlib's object tracking. Give it a bounding box and it will try and track it using correlation.\n\nhttps://www.pyimagesearch.com/2018/10/22/object-tracking-with-dlib/",
    "715936": "Opencv has very good tracker and multitracker"
  },
  "source": "meta"
}