{
  "id": 121656,
  "title": "First Folder compressed for Starters",
  "url": "/competitions/deepfake-detection-challenge/discussion/121656",
  "author_name": "",
  "post_date": "2019-12-14T15:54:08.622925100Z",
  "votes": 2,
  "comment_count": 3,
  "views": 0,
  "content": "<p>For starters I created a Dataset consisting of the faces in the first frame of each video in folder 00 of training data.</p>\n\n<p>For face detection I used: <a href=\"https://github.com/ipazc/mtcnn\">MTCNN</a> and boxes were enlarged by 80 pixels in width and height.</p>\n\n<p>Dataset:\n<a href=\"https://www.kaggle.com/jpbremer/deepfake00\">https://www.kaggle.com/jpbremer/deepfake00</a></p>\n\n<p>The original metadata is provided. \nImage names end by index ( _0, _1, ...) indicating how many images are detected per frame.</p>",
  "messages": [
    {
      "id": "695113",
      "postDate": "12/14/2019 15:54:08",
      "content": "<p>For starters I created a Dataset consisting of the faces in the first frame of each video in folder 00 of training data.</p>\n\n<p>For face detection I used: <a href=\"https://github.com/ipazc/mtcnn\">MTCNN</a> and boxes were enlarged by 80 pixels in width and height.</p>\n\n<p>Dataset:\n<a href=\"https://www.kaggle.com/jpbremer/deepfake00\">https://www.kaggle.com/jpbremer/deepfake00</a></p>\n\n<p>The original metadata is provided. \nImage names end by index ( _0, _1, ...) indicating how many images are detected per frame.</p>",
      "rawMarkdown": "For starters I created a Dataset consisting of the faces in the first frame of each video in folder 00 of training data.\n\nFor face detection I used: [MTCNN](https://github.com/ipazc/mtcnn) and boxes were enlarged by 80 pixels in width and height.\n\nDataset:\n[https://www.kaggle.com/jpbremer/deepfake00](https://www.kaggle.com/jpbremer/deepfake00)\n\nThe original metadata is provided. \nImage names end by index ( _0, _1, ...) indicating how many images are detected per frame.",
      "votes": null
    },
    {
      "id": "695121",
      "postDate": "12/14/2019 16:08:27",
      "content": "<p>Looks great! How did you deal with videos with two people in them?</p>",
      "rawMarkdown": "Looks great! How did you deal with videos with two people in them?",
      "votes": null
    },
    {
      "id": "695136",
      "postDate": "12/14/2019 16:40:12",
      "content": "<p>If there are multiple people in a video MTCNN detects multiple faces per image indicated by the Index in this dataset.</p>",
      "rawMarkdown": "If there are multiple people in a video MTCNN detects multiple faces per image indicated by the Index in this dataset.",
      "votes": null
    },
    {
      "id": "695176",
      "postDate": "12/14/2019 17:44:44",
      "content": "<p>Thank very much. This is very helpful. Can you upload the code used so we can also generate the test data as well?</p>",
      "rawMarkdown": "Thank very much. This is very helpful. Can you upload the code used so we can also generate the test data as well?",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 695121,
      "author_name": "sdoria",
      "author_url": "",
      "post_date": "12/14/2019 16:08:27",
      "content": "<p>Looks great! How did you deal with videos with two people in them?</p>",
      "votes": null,
      "replies": [
        {
          "id": 695136,
          "author_name": "jpbremer",
          "author_url": "",
          "post_date": "12/14/2019 16:40:12",
          "content": "<p>If there are multiple people in a video MTCNN detects multiple faces per image indicated by the Index in this dataset.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 695176,
      "author_name": "unkownhihi",
      "author_url": "",
      "post_date": "12/14/2019 17:44:44",
      "content": "<p>Thank very much. This is very helpful. Can you upload the code used so we can also generate the test data as well?</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "695113": "For starters I created a Dataset consisting of the faces in the first frame of each video in folder 00 of training data.\n\nFor face detection I used: [MTCNN](https://github.com/ipazc/mtcnn) and boxes were enlarged by 80 pixels in width and height.\n\nDataset:\n[https://www.kaggle.com/jpbremer/deepfake00](https://www.kaggle.com/jpbremer/deepfake00)\n\nThe original metadata is provided. \nImage names end by index ( _0, _1, ...) indicating how many images are detected per frame.",
    "695121": "Looks great! How did you deal with videos with two people in them?",
    "695136": "If there are multiple people in a video MTCNN detects multiple faces per image indicated by the Index in this dataset.",
    "695176": "Thank very much. This is very helpful. Can you upload the code used so we can also generate the test data as well?"
  },
  "source": "meta"
}