{
  "id": 5879,
  "title": "Cropping",
  "url": "/competitions/dogs-vs-cats/discussion/5879",
  "author_name": "",
  "post_date": "2013-09-25T20:39:42.477Z",
  "votes": null,
  "comment_count": 5,
  "views": 3480,
  "content": "<p>It sort of sucks that the images are not the same dimension. This is probably a dumb question but are their algorithms that take variable number of features or is it assume we have to come up with a method for cropping also ?</p>",
  "messages": [
    {
      "id": "31568",
      "postDate": "09/25/2013 20:39:42",
      "content": "<p>It sort of sucks that the images are not the same dimension. This is probably a dumb question but are their algorithms that take variable number of features or is it assume we have to come up with a method for cropping also ?</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "31569",
      "postDate": "09/25/2013 21:25:34",
      "content": "<p>You should find features independent of image size. In normal life, all images are not the same size...</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "31575",
      "postDate": "09/26/2013 01:47:36",
      "content": "<p>You could try re-sizing and stretching the images to a fixed resolution. This will give you a set number of pixels.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "31648",
      "postDate": "09/27/2013 08:12:40",
      "content": "<p>Somewhat related : maybe a face tracking algo can help</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "31674",
      "postDate": "09/27/2013 21:46:56",
      "content": "<p>[quote=lucy24;31569]</p>\n<p>You should find features independent of image size. In normal life, all images are not the same size...</p>\n<p>[/quote]</p>\n<p>in normal life? what's that?&nbsp;</p>\n<p>all kidding aside, you stated the obvious and didn't answer the question</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "35728",
      "postDate": "12/03/2013 04:28:34",
      "content": "<p>Not sure about real life but in object detection in &nbsp;videos, all frames are of same size</p>\n<p>Also, most libraries assume or want the images to be same size to work properly</p>\n<p>So, I understand that in this case we have to make all the images to be of same size before we can train our classifier</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 31569,
      "author_name": "amtousch",
      "author_url": "",
      "post_date": "09/25/2013 21:25:34",
      "content": "<p>You should find features independent of image size. In normal life, all images are not the same size...</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 31575,
      "author_name": "primaryobjects",
      "author_url": "",
      "post_date": "09/26/2013 01:47:36",
      "content": "<p>You could try re-sizing and stretching the images to a fixed resolution. This will give you a set number of pixels.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 31648,
      "author_name": "rudikruger",
      "author_url": "",
      "post_date": "09/27/2013 08:12:40",
      "content": "<p>Somewhat related : maybe a face tracking algo can help</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 31674,
      "author_name": "jhuling",
      "author_url": "",
      "post_date": "09/27/2013 21:46:56",
      "content": "<p>[quote=lucy24;31569]</p>\n<p>You should find features independent of image size. In normal life, all images are not the same size...</p>\n<p>[/quote]</p>\n<p>in normal life? what's that?&nbsp;</p>\n<p>all kidding aside, you stated the obvious and didn't answer the question</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 35728,
      "author_name": "nvedia",
      "author_url": "",
      "post_date": "12/03/2013 04:28:34",
      "content": "<p>Not sure about real life but in object detection in &nbsp;videos, all frames are of same size</p>\n<p>Also, most libraries assume or want the images to be same size to work properly</p>\n<p>So, I understand that in this case we have to make all the images to be of same size before we can train our classifier</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "31568": "",
    "31569": "",
    "31575": "",
    "31648": "",
    "31674": "",
    "35728": ""
  },
  "source": "meta"
}