{
  "id": 242706,
  "title": "Study Level Images & Single images",
  "url": "/competitions/siim-covid19-detection/discussion/242706",
  "author_name": "Izzy Adesanya",
  "post_date": "2021-05-30T10:15:11.814000",
  "votes": 2,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Please help me clarify a couple of questions regarding the Study level and Single images.</p>\n<p>Here are the things I know:</p>\n<ol>\n<li>Studies either contain a single image (majority this) or multiple images.</li>\n<li>All of the images in the studies are also found in the image-level data.</li>\n<li>Images in the study can have different labels.</li>\n</ol>\n<p>Questions:</p>\n<ol>\n<li>In the case of multiple images in a study, how do we handle them while generating data for the model? Do we discard the images that don't have any labels or bounding box info for them?</li>\n<li>Do we make study-level predictions or image-level predictions? (I suppose it should be study level since we train on study level images)</li>\n<li>When images in a study have different labels, how do we handle that scenario?</li>\n</ol>\n<p>Feel free to point out if I stated anything wrong here.<br>\nTIA 😊</p>",
  "messages": [
    {
      "id": 1328515,
      "postDate": "2021-05-30T10:15:11.813Z",
      "content": "<p>Please help me clarify a couple of questions regarding the Study level and Single images.</p>\n<p>Here are the things I know:</p>\n<ol>\n<li>Studies either contain a single image (majority this) or multiple images.</li>\n<li>All of the images in the studies are also found in the image-level data.</li>\n<li>Images in the study can have different labels.</li>\n</ol>\n<p>Questions:</p>\n<ol>\n<li>In the case of multiple images in a study, how do we handle them while generating data for the model? Do we discard the images that don't have any labels or bounding box info for them?</li>\n<li>Do we make study-level predictions or image-level predictions? (I suppose it should be study level since we train on study level images)</li>\n<li>When images in a study have different labels, how do we handle that scenario?</li>\n</ol>\n<p>Feel free to point out if I stated anything wrong here.<br>\nTIA 😊</p>",
      "rawMarkdown": "Please help me clarify a couple of questions regarding the Study level and Single images.\n\nHere are the things I know:\n\n1. Studies either contain a single image (majority this) or multiple images.\n2. All of the images in the studies are also found in the image-level data.\n3. Images in the study can have different labels.\n\nQuestions:\n1. In the case of multiple images in a study, how do we handle them while generating data for the model? Do we discard the images that don't have any labels or bounding box info for them?\n2. Do we make study-level predictions or image-level predictions? (I suppose it should be study level since we train on study level images)\n3. When images in a study have different labels, how do we handle that scenario?\n\nFeel free to point out if I stated anything wrong here.\nTIA 😊",
      "votes": 2
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "1328515": "Please help me clarify a couple of questions regarding the Study level and Single images.\n\nHere are the things I know:\n\n1. Studies either contain a single image (majority this) or multiple images.\n2. All of the images in the studies are also found in the image-level data.\n3. Images in the study can have different labels.\n\nQuestions:\n1. In the case of multiple images in a study, how do we handle them while generating data for the model? Do we discard the images that don't have any labels or bounding box info for them?\n2. Do we make study-level predictions or image-level predictions? (I suppose it should be study level since we train on study level images)\n3. When images in a study have different labels, how do we handle that scenario?\n\nFeel free to point out if I stated anything wrong here.\nTIA 😊"
  }
}