{
  "id": 1210,
  "title": "A question on color correlation",
  "url": "/competitions/GestureChallenge/discussion/1210",
  "author_name": "",
  "post_date": "2012-01-02T01:11:04.937Z",
  "votes": null,
  "comment_count": 1,
  "views": 1462,
  "content": "<p>Consider the devel03 training files.</p>\r\n<p>&nbsp;</p>\r\n<p>M_1 shows the actor wearing a black bib which covers most of his body. Test videos of this gesture have him wearing a white T-shirt.</p>\r\n<p>In M_2 he has the bib on his lap, and is wearing a patterned shirt. Test videos of this gesture have him wearing a white T-shirt.</p>\r\n<p>In M_3 .. M_6 there is no bib but there is a shirt. Test videos of these gestures have him wearing a white T-shirt.</p>\r\n<p>In M_7 .. M_8 he is wearing a white T-shirt</p>\r\n<p>&nbsp;</p>\r\n<p>Just to be clear, you would like a system which trains on color images where the actor is wearing one set of clothing, and then interprets gestures when the clothing is significantly different.</p>\r\n<p>&nbsp;</p>\r\n<p>This is correct, yes?</p>",
  "messages": [
    {
      "id": "7609",
      "postDate": "01/02/2012 01:11:04",
      "content": "<p>Consider the devel03 training files.</p>\r\n<p>&nbsp;</p>\r\n<p>M_1 shows the actor wearing a black bib which covers most of his body. Test videos of this gesture have him wearing a white T-shirt.</p>\r\n<p>In M_2 he has the bib on his lap, and is wearing a patterned shirt. Test videos of this gesture have him wearing a white T-shirt.</p>\r\n<p>In M_3 .. M_6 there is no bib but there is a shirt. Test videos of these gestures have him wearing a white T-shirt.</p>\r\n<p>In M_7 .. M_8 he is wearing a white T-shirt</p>\r\n<p>&nbsp;</p>\r\n<p>Just to be clear, you would like a system which trains on color images where the actor is wearing one set of clothing, and then interprets gestures when the clothing is significantly different.</p>\r\n<p>&nbsp;</p>\r\n<p>This is correct, yes?</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "7641",
      "postDate": "01/03/2012 19:32:33",
      "content": "<p>We have some of the same users in the development and validation data, but you should not count of necessarily getting the same users in the final evaluation data. For sure you will get none of the same gesture vocabularies. Each task (corresponding to a\r\n data batch) has a single user and a particular gesture vocabulary and you get one example of each gesture. Some time the user takes off or on a sweater in the middle of the batch. Some time the lighting changes (e.g. the sun sets, the user turns on the light).\r\n This is part of the problem.</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 7641,
      "author_name": "iguyon",
      "author_url": "",
      "post_date": "01/03/2012 19:32:33",
      "content": "<p>We have some of the same users in the development and validation data, but you should not count of necessarily getting the same users in the final evaluation data. For sure you will get none of the same gesture vocabularies. Each task (corresponding to a\r\n data batch) has a single user and a particular gesture vocabulary and you get one example of each gesture. Some time the user takes off or on a sweater in the middle of the batch. Some time the lighting changes (e.g. the sun sets, the user turns on the light).\r\n This is part of the problem.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "7609": "",
    "7641": ""
  },
  "source": "meta"
}