{
  "id": 40058,
  "title": "Ground Truth",
  "url": "/competitions/passenger-screening-algorithm-challenge/discussion/40058",
  "author_name": "",
  "post_date": "2017-09-26T22:34:24.088706300Z",
  "votes": null,
  "comment_count": 3,
  "views": 0,
  "content": "<p>For training purposes, it is required to generate the Ground Truths (GTs) of the images, that is, label the areas (or pixels) that are considered as threats and the areas that are considered clean. Does anyone know any fast way to create these label images? Since there are tones of images, this process can take for ever.</p>\n\n<p>Thank you.</p>",
  "messages": [
    {
      "id": "224596",
      "postDate": "09/26/2017 22:34:24",
      "content": "<p>For training purposes, it is required to generate the Ground Truths (GTs) of the images, that is, label the areas (or pixels) that are considered as threats and the areas that are considered clean. Does anyone know any fast way to create these label images? Since there are tones of images, this process can take for ever.</p>\n\n<p>Thank you.</p>",
      "rawMarkdown": "For training purposes, it is required to generate the Ground Truths (GTs) of the images, that is, label the areas (or pixels) that are considered as threats and the areas that are considered clean. Does anyone know any fast way to create these label images? Since there are tones of images, this process can take for ever.\n\nThank you.",
      "votes": null
    },
    {
      "id": "224727",
      "postDate": "09/27/2017 13:48:28",
      "content": "<p>The ground truth for the training data is contained in stage1_labels.csv</p>",
      "rawMarkdown": "The ground truth for the training data is contained in stage1_labels.csv",
      "votes": null
    },
    {
      "id": "224774",
      "postDate": "09/27/2017 16:24:46",
      "content": "<p>Thank you. I know that stage1_labels.csv contains information about if there is a threat in each zone of the body of each subject or not, but there is no information about the shape and type of threat. My question is about creating a label image, indicating the pixels that correspond to a threat and the ones that not (even different types of threats may be defined with different labels).</p>",
      "rawMarkdown": "Thank you. I know that stage1_labels.csv contains information about if there is a threat in each zone of the body of each subject or not, but there is no information about the shape and type of threat. My question is about creating a label image, indicating the pixels that correspond to a threat and the ones that not (even different types of threats may be defined with different labels).",
      "votes": null
    },
    {
      "id": "225565",
      "postDate": "09/29/2017 14:19:38",
      "content": "<p>That's the challenge - and it's a reasonable challenge in the sense that in real life the types of weapons and ways to conceal them is pretty open-ended.</p>",
      "rawMarkdown": "That's the challenge - and it's a reasonable challenge in the sense that in real life the types of weapons and ways to conceal them is pretty open-ended.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 224727,
      "author_name": "wcukierski",
      "author_url": "",
      "post_date": "09/27/2017 13:48:28",
      "content": "<p>The ground truth for the training data is contained in stage1_labels.csv</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 224774,
      "author_name": "juanheredia",
      "author_url": "",
      "post_date": "09/27/2017 16:24:46",
      "content": "<p>Thank you. I know that stage1_labels.csv contains information about if there is a threat in each zone of the body of each subject or not, but there is no information about the shape and type of threat. My question is about creating a label image, indicating the pixels that correspond to a threat and the ones that not (even different types of threats may be defined with different labels).</p>",
      "votes": null,
      "replies": [
        {
          "id": 225565,
          "author_name": "tothink",
          "author_url": "",
          "post_date": "09/29/2017 14:19:38",
          "content": "<p>That's the challenge - and it's a reasonable challenge in the sense that in real life the types of weapons and ways to conceal them is pretty open-ended.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "224596": "For training purposes, it is required to generate the Ground Truths (GTs) of the images, that is, label the areas (or pixels) that are considered as threats and the areas that are considered clean. Does anyone know any fast way to create these label images? Since there are tones of images, this process can take for ever.\n\nThank you.",
    "224727": "The ground truth for the training data is contained in stage1_labels.csv",
    "224774": "Thank you. I know that stage1_labels.csv contains information about if there is a threat in each zone of the body of each subject or not, but there is no information about the shape and type of threat. My question is about creating a label image, indicating the pixels that correspond to a threat and the ones that not (even different types of threats may be defined with different labels).",
    "225565": "That's the challenge - and it's a reasonable challenge in the sense that in real life the types of weapons and ways to conceal them is pretty open-ended."
  },
  "source": "meta"
}