{
  "id": 100674,
  "title": "Training and testing data size issues",
  "url": "/competitions/open-images-2019-object-detection/discussion/100674",
  "author_name": "",
  "post_date": "2019-07-20T04:02:07.252091400Z",
  "votes": 1,
  "comment_count": 1,
  "views": 0,
  "content": "<p>If I train the model and convert the image to 500 *500.  Scale the image in the training set\nSo when I do the test, should I also convert the picture into 500*500 for the test?\nHow does this affect the outcome?\nOr just use the test set data without any processing?\nI would be grateful if anyone could answer</p>",
  "messages": [
    {
      "id": "580370",
      "postDate": "07/20/2019 04:02:07",
      "content": "<p>If I train the model and convert the image to 500 *500.  Scale the image in the training set\nSo when I do the test, should I also convert the picture into 500*500 for the test?\nHow does this affect the outcome?\nOr just use the test set data without any processing?\nI would be grateful if anyone could answer</p>",
      "rawMarkdown": "If I train the model and convert the image to 500 *500.  Scale the image in the training set\nSo when I do the test, should I also convert the picture into 500*500 for the test?\nHow does this affect the outcome?\nOr just use the test set data without any processing?\nI would be grateful if anyone could answer",
      "votes": null
    },
    {
      "id": "580425",
      "postDate": "07/20/2019 06:14:01",
      "content": "<p>hello <a href=\"/hsykaggle\">@hsykaggle</a> ,</p>\n\n<p>As your Model is trained on a specific format of Data , it is advisable to have your Test data converted to same format so that the training can be applied conclusively on the Test data to predict the outcome.</p>\n\n<p>O</p>",
      "rawMarkdown": "hello @hsykaggle ,\n\nAs your Model is trained on a specific format of Data , it is advisable to have your Test data converted to same format so that the training can be applied conclusively on the Test data to predict the outcome.\n\nO",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 580425,
      "author_name": "prmohanty",
      "author_url": "",
      "post_date": "07/20/2019 06:14:01",
      "content": "<p>hello <a href=\"/hsykaggle\">@hsykaggle</a> ,</p>\n\n<p>As your Model is trained on a specific format of Data , it is advisable to have your Test data converted to same format so that the training can be applied conclusively on the Test data to predict the outcome.</p>\n\n<p>O</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "580370": "If I train the model and convert the image to 500 *500.  Scale the image in the training set\nSo when I do the test, should I also convert the picture into 500*500 for the test?\nHow does this affect the outcome?\nOr just use the test set data without any processing?\nI would be grateful if anyone could answer",
    "580425": "hello @hsykaggle ,\n\nAs your Model is trained on a specific format of Data , it is advisable to have your Test data converted to same format so that the training can be applied conclusively on the Test data to predict the outcome.\n\nO"
  },
  "source": "meta"
}