{
  "id": 253483,
  "title": "Formulation of submission.csv file",
  "url": "/competitions/siim-covid19-detection/discussion/253483",
  "author_name": "",
  "post_date": "2021-07-16T21:19:44.452999100Z",
  "votes": 1,
  "comment_count": 1,
  "views": 0,
  "content": "<p>The submission file contains the id_study and id_image data.<br>\nIn id_image you need to save the data as follows:<br>\n*name1_image,opacity 0.416 611 458 1152 1203 opacity 0.383 1874 722 2394 1703<br>\nname2_image, none 1 0 0 1 1<br>\n…<br>\nbeing, the confidence level and the bounding box coordinates, this for each opacity found in each image</p>\n<p>In id_study, a study can contain multiple images.<br>\nThe data to be included in csv would be a classification for 4 classes (negative, typical, atypical and undetermined) with the result for each output (confidence level) and 0 0 1 1 for the bounding boxes?<br>\nWhat about studies that contain more than one image?</p>",
  "messages": [
    {
      "id": "1390624",
      "postDate": "07/16/2021 21:19:44",
      "content": "<p>The submission file contains the id_study and id_image data.<br>\nIn id_image you need to save the data as follows:<br>\n*name1_image,opacity 0.416 611 458 1152 1203 opacity 0.383 1874 722 2394 1703<br>\nname2_image, none 1 0 0 1 1<br>\n…<br>\nbeing, the confidence level and the bounding box coordinates, this for each opacity found in each image</p>\n<p>In id_study, a study can contain multiple images.<br>\nThe data to be included in csv would be a classification for 4 classes (negative, typical, atypical and undetermined) with the result for each output (confidence level) and 0 0 1 1 for the bounding boxes?<br>\nWhat about studies that contain more than one image?</p>",
      "rawMarkdown": "The submission file contains the id_study and id_image data.\nIn id_image you need to save the data as follows:\n*name1_image,opacity 0.416 611 458 1152 1203 opacity 0.383 1874 722 2394 1703\nname2_image, none 1 0 0 1 1\n...\nbeing, the confidence level and the bounding box coordinates, this for each opacity found in each image\n\n\nIn id_study, a study can contain multiple images.\nThe data to be included in csv would be a classification for 4 classes (negative, typical, atypical and undetermined) with the result for each output (confidence level) and 0 0 1 1 for the bounding boxes?\nWhat about studies that contain more than one image?",
      "votes": null
    },
    {
      "id": "1390741",
      "postDate": "07/17/2021 04:00:55",
      "content": "<p>Hi.</p>\n<p>You are right about _image predictions. Each image can have one bounding box, multiple bounding boxes, or none. In _image case you would save your predictions such as:<br>\nname1_image,opacity 0.416 611 458 1152 1203<br>\nname2_image,opacity 0.416 611 458 1152 1203 opacity 0.383 1874 722 2394 1703<br>\nname3_image,none 1 0 0 1 1</p>\n<p>About _study predictions, yes, some studies have multiple images in them. In this case it is up to you on how to use your predictions. What you might do is to do classify each image in a study and take mean of your predictions, where the result (mean) is going to be your final prediction, like<br>\nname1_study,indeterminate 0.97856 0 0 1 1 atypical 0.54365 0 0 1 1<br>\nOr you can randomly choose a file in a study and use it as a final _study prediction (@Philipp Rajah Moura Srivastava suggested it in the same question that I asked here:<br>\n<a href=\"https://www.kaggle.com/c/siim-covid19-detection/discussion/251807\" target=\"_blank\">https://www.kaggle.com/c/siim-covid19-detection/discussion/251807</a>)</p>",
      "rawMarkdown": "Hi.\n\nYou are right about _image predictions. Each image can have one bounding box, multiple bounding boxes, or none. In _image case you would save your predictions such as:\nname1_image,opacity 0.416 611 458 1152 1203\nname2_image,opacity 0.416 611 458 1152 1203 opacity 0.383 1874 722 2394 1703\nname3_image,none 1 0 0 1 1\n\nAbout _study predictions, yes, some studies have multiple images in them. In this case it is up to you on how to use your predictions. What you might do is to do classify each image in a study and take mean of your predictions, where the result (mean) is going to be your final prediction, like\nname1_study,indeterminate 0.97856 0 0 1 1 atypical 0.54365 0 0 1 1\nOr you can randomly choose a file in a study and use it as a final _study prediction (@Philipp Rajah Moura Srivastava suggested it in the same question that I asked here:\nhttps://www.kaggle.com/c/siim-covid19-detection/discussion/251807)",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1390741,
      "author_name": "antonzv",
      "author_url": "",
      "post_date": "07/17/2021 04:00:55",
      "content": "<p>Hi.</p>\n<p>You are right about _image predictions. Each image can have one bounding box, multiple bounding boxes, or none. In _image case you would save your predictions such as:<br>\nname1_image,opacity 0.416 611 458 1152 1203<br>\nname2_image,opacity 0.416 611 458 1152 1203 opacity 0.383 1874 722 2394 1703<br>\nname3_image,none 1 0 0 1 1</p>\n<p>About _study predictions, yes, some studies have multiple images in them. In this case it is up to you on how to use your predictions. What you might do is to do classify each image in a study and take mean of your predictions, where the result (mean) is going to be your final prediction, like<br>\nname1_study,indeterminate 0.97856 0 0 1 1 atypical 0.54365 0 0 1 1<br>\nOr you can randomly choose a file in a study and use it as a final _study prediction (@Philipp Rajah Moura Srivastava suggested it in the same question that I asked here:<br>\n<a href=\"https://www.kaggle.com/c/siim-covid19-detection/discussion/251807\" target=\"_blank\">https://www.kaggle.com/c/siim-covid19-detection/discussion/251807</a>)</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1390624": "The submission file contains the id_study and id_image data.\nIn id_image you need to save the data as follows:\n*name1_image,opacity 0.416 611 458 1152 1203 opacity 0.383 1874 722 2394 1703\nname2_image, none 1 0 0 1 1\n...\nbeing, the confidence level and the bounding box coordinates, this for each opacity found in each image\n\n\nIn id_study, a study can contain multiple images.\nThe data to be included in csv would be a classification for 4 classes (negative, typical, atypical and undetermined) with the result for each output (confidence level) and 0 0 1 1 for the bounding boxes?\nWhat about studies that contain more than one image?",
    "1390741": "Hi.\n\nYou are right about _image predictions. Each image can have one bounding box, multiple bounding boxes, or none. In _image case you would save your predictions such as:\nname1_image,opacity 0.416 611 458 1152 1203\nname2_image,opacity 0.416 611 458 1152 1203 opacity 0.383 1874 722 2394 1703\nname3_image,none 1 0 0 1 1\n\nAbout _study predictions, yes, some studies have multiple images in them. In this case it is up to you on how to use your predictions. What you might do is to do classify each image in a study and take mean of your predictions, where the result (mean) is going to be your final prediction, like\nname1_study,indeterminate 0.97856 0 0 1 1 atypical 0.54365 0 0 1 1\nOr you can randomly choose a file in a study and use it as a final _study prediction (@Philipp Rajah Moura Srivastava suggested it in the same question that I asked here:\nhttps://www.kaggle.com/c/siim-covid19-detection/discussion/251807)"
  },
  "source": "meta"
}