{
  "id": 140495,
  "title": "Extracted Face Dataset",
  "url": "/competitions/deepfake-detection-challenge/discussion/140495",
  "author_name": "Ken Ho",
  "post_date": "2020-04-02T02:52:56.857000",
  "votes": 0,
  "comment_count": 4,
  "views": 0,
  "content": "<p>Data is prepared in reference to this kernel, credit to @unkownhihi :\n<a href=\"https://www.kaggle.com/unkownhihi/dfdc-lrcn-inference\">https://www.kaggle.com/unkownhihi/dfdc-lrcn-inference</a></p>\n\n<p>One png is from 10 faces in a row from the same video.</p>\n\n<p><a href=\"https://www.kaggle.com/kenho211/dfdc-face1\">Part 1</a>\n<a href=\"https://www.kaggle.com/kenho211/dfdc-face2\">Part 2</a>\n<a href=\"https://www.kaggle.com/kenho211/dfdc-face3\">Part 3</a>\n<a href=\"https://www.kaggle.com/kenho211/dfdc-face4\">Part 4</a>\n<a href=\"https://www.kaggle.com/kenho211/dfdc-face5\">Part 5</a></p>\n\n<p>Note: agorzaxkul.png in part 25 is corrupted.</p>",
  "messages": [
    {
      "id": 794698,
      "postDate": "2020-04-02T02:52:56.857Z",
      "content": "<p>Data is prepared in reference to this kernel, credit to @unkownhihi :\n<a href=\"https://www.kaggle.com/unkownhihi/dfdc-lrcn-inference\">https://www.kaggle.com/unkownhihi/dfdc-lrcn-inference</a></p>\n\n<p>One png is from 10 faces in a row from the same video.</p>\n\n<p><a href=\"https://www.kaggle.com/kenho211/dfdc-face1\">Part 1</a>\n<a href=\"https://www.kaggle.com/kenho211/dfdc-face2\">Part 2</a>\n<a href=\"https://www.kaggle.com/kenho211/dfdc-face3\">Part 3</a>\n<a href=\"https://www.kaggle.com/kenho211/dfdc-face4\">Part 4</a>\n<a href=\"https://www.kaggle.com/kenho211/dfdc-face5\">Part 5</a></p>\n\n<p>Note: agorzaxkul.png in part 25 is corrupted.</p>",
      "rawMarkdown": "Data is prepared in reference to this kernel, credit to @unkownhihi :\nhttps://www.kaggle.com/unkownhihi/dfdc-lrcn-inference\n\nOne png is from 10 faces in a row from the same video.\n\n[Part 1](https://www.kaggle.com/kenho211/dfdc-face1)\n[Part 2](https://www.kaggle.com/kenho211/dfdc-face2)\n[Part 3](https://www.kaggle.com/kenho211/dfdc-face3)\n[Part 4](https://www.kaggle.com/kenho211/dfdc-face4)\n[Part 5](https://www.kaggle.com/kenho211/dfdc-face5)\n\nNote: agorzaxkul.png in part 25 is corrupted."
    },
    {
      "id": 796868,
      "postDate": "2020-04-04T02:38:41.677Z",
      "rawMarkdown": "",
      "isDeleted": true
    },
    {
      "id": 796866,
      "postDate": "2020-04-04T02:37:57.143Z",
      "content": "<p>Are they complete? How you made this dataset? Why you tile these png in a video together? Thanks!</p>",
      "rawMarkdown": "Are they complete? How you made this dataset? Why you tile these png in a video together? Thanks!\n",
      "isDeleted": true,
      "replies": [
        {
          "id": 798527,
          "postDate": "2020-04-05T15:23:03.487Z",
          "content": "<p>They are complete. There is an image for each video. For the script to produce it, it mainly follows <a href=\"https://www.kaggle.com/unkownhihi/dfdc-lrcn-inference\">https://www.kaggle.com/unkownhihi/dfdc-lrcn-inference</a>\nIt is tiled to 10 images horizontally as the input shape of TimeDistributed layer is (10, 240,240,3)</p>",
          "rawMarkdown": "They are complete. There is an image for each video. For the script to produce it, it mainly follows https://www.kaggle.com/unkownhihi/dfdc-lrcn-inference\nIt is tiled to 10 images horizontally as the input shape of TimeDistributed layer is (10, 240,240,3)"
        },
        {
          "id": 798831,
          "postDate": "2020-04-05T22:20:12.600Z",
          "content": "<p>Thanks!</p>",
          "rawMarkdown": "Thanks!",
          "isDeleted": true
        }
      ]
    }
  ],
  "comments": [
    {
      "id": 796868,
      "author_name": "",
      "author_url": "",
      "post_date": "2020-04-04T02:38:41.677000",
      "content": "",
      "votes": 0,
      "replies": []
    },
    {
      "id": 796866,
      "author_name": "",
      "author_url": "",
      "post_date": "2020-04-04T02:37:57.143000",
      "content": "<p>Are they complete? How you made this dataset? Why you tile these png in a video together? Thanks!</p>",
      "votes": 0,
      "replies": [
        {
          "id": 798527,
          "author_name": "Ken Ho",
          "author_url": "",
          "post_date": "2020-04-05T15:23:03.487000",
          "content": "<p>They are complete. There is an image for each video. For the script to produce it, it mainly follows <a href=\"https://www.kaggle.com/unkownhihi/dfdc-lrcn-inference\">https://www.kaggle.com/unkownhihi/dfdc-lrcn-inference</a>\nIt is tiled to 10 images horizontally as the input shape of TimeDistributed layer is (10, 240,240,3)</p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 798831,
          "author_name": "",
          "author_url": "",
          "post_date": "2020-04-05T22:20:12.600000",
          "content": "<p>Thanks!</p>",
          "votes": 0,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "794698": "Data is prepared in reference to this kernel, credit to @unkownhihi :\nhttps://www.kaggle.com/unkownhihi/dfdc-lrcn-inference\n\nOne png is from 10 faces in a row from the same video.\n\n[Part 1](https://www.kaggle.com/kenho211/dfdc-face1)\n[Part 2](https://www.kaggle.com/kenho211/dfdc-face2)\n[Part 3](https://www.kaggle.com/kenho211/dfdc-face3)\n[Part 4](https://www.kaggle.com/kenho211/dfdc-face4)\n[Part 5](https://www.kaggle.com/kenho211/dfdc-face5)\n\nNote: agorzaxkul.png in part 25 is corrupted.",
    "796868": "",
    "796866": "Are they complete? How you made this dataset? Why you tile these png in a video together? Thanks!\n"
  }
}