{
  "id": 245109,
  "title": "Question about abnormal object detection.",
  "url": "/competitions/siim-covid19-detection/discussion/245109",
  "author_name": "David Roberts",
  "post_date": "2021-06-09T17:23:35.750000",
  "votes": 0,
  "comment_count": 0,
  "views": 0,
  "content": "<p>I understand how classification with a CNN works on the entire image, but I am not sure how the bounding box part fits in conceptually.</p>\n<p>Is the common basic strategy to extract all the ROIs from the training image BB's and use them as training images for object detection on the test images? </p>",
  "messages": [
    {
      "id": 1342729,
      "postDate": "2021-06-09T17:23:35.750Z",
      "content": "<p>I understand how classification with a CNN works on the entire image, but I am not sure how the bounding box part fits in conceptually.</p>\n<p>Is the common basic strategy to extract all the ROIs from the training image BB's and use them as training images for object detection on the test images? </p>",
      "rawMarkdown": "I understand how classification with a CNN works on the entire image, but I am not sure how the bounding box part fits in conceptually.\n\nIs the common basic strategy to extract all the ROIs from the training image BB's and use them as training images for object detection on the test images? "
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "1342729": "I understand how classification with a CNN works on the entire image, but I am not sure how the bounding box part fits in conceptually.\n\nIs the common basic strategy to extract all the ROIs from the training image BB's and use them as training images for object detection on the test images? "
  }
}