{
  "id": 312158,
  "title": "Can computer vision algorithms be used to detect characteristics?",
  "url": "/competitions/happy-whale-and-dolphin/discussion/312158",
  "author_name": "",
  "post_date": "2022-03-10T17:19:38.451944200Z",
  "votes": 2,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Hi Kagglers, </p>\n<p>I was wondering if computer vision algorithms could help identify the pixels present in images. I have used CV algorithms in classification of images and training the algos based on image data. We have image data in this competition. Would any computer vision would be useful for detecting unique pixels in images as characteristics?</p>",
  "messages": [
    {
      "id": "1718270",
      "postDate": "03/10/2022 17:19:38",
      "content": "<p>Hi Kagglers, </p>\n<p>I was wondering if computer vision algorithms could help identify the pixels present in images. I have used CV algorithms in classification of images and training the algos based on image data. We have image data in this competition. Would any computer vision would be useful for detecting unique pixels in images as characteristics?</p>",
      "rawMarkdown": "Hi Kagglers, \n\nI was wondering if computer vision algorithms could help identify the pixels present in images. I have used CV algorithms in classification of images and training the algos based on image data. We have image data in this competition. Would any computer vision would be useful for detecting unique pixels in images as characteristics?",
      "votes": null
    },
    {
      "id": "1718855",
      "postDate": "03/11/2022 07:58:25",
      "content": "<p>Usually \"unique\" pixels are not characteristics, conv networks work with windows (area around pixel, not only one) and high levels operate areas with size close one to whole image. Also, we have keypoint features like SURF/SIFT/LIFT and others, here is one pixel yes, but also with context, not just \"only one pixel\"</p>",
      "rawMarkdown": "Usually \"unique\" pixels are not characteristics, conv networks work with windows (area around pixel, not only one) and high levels operate areas with size close one to whole image. Also, we have keypoint features like SURF/SIFT/LIFT and others, here is one pixel yes, but also with context, not just \"only one pixel\"",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1718855,
      "author_name": "kwentar",
      "author_url": "",
      "post_date": "03/11/2022 07:58:25",
      "content": "<p>Usually \"unique\" pixels are not characteristics, conv networks work with windows (area around pixel, not only one) and high levels operate areas with size close one to whole image. Also, we have keypoint features like SURF/SIFT/LIFT and others, here is one pixel yes, but also with context, not just \"only one pixel\"</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1718270": "Hi Kagglers, \n\nI was wondering if computer vision algorithms could help identify the pixels present in images. I have used CV algorithms in classification of images and training the algos based on image data. We have image data in this competition. Would any computer vision would be useful for detecting unique pixels in images as characteristics?",
    "1718855": "Usually \"unique\" pixels are not characteristics, conv networks work with windows (area around pixel, not only one) and high levels operate areas with size close one to whole image. Also, we have keypoint features like SURF/SIFT/LIFT and others, here is one pixel yes, but also with context, not just \"only one pixel\""
  },
  "source": "meta"
}