{
  "id": 170500,
  "title": "Submission error",
  "url": "/competitions/global-wheat-detection/discussion/170500",
  "author_name": "",
  "post_date": "2020-07-28T02:11:04.607870400Z",
  "votes": 1,
  "comment_count": 2,
  "views": 0,
  "content": "<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2976720%2F4b93f043cd4aa12d3b37ab35a3ae8a9c%2FQQ20200728100726.jpg?generation=1595902096112245&amp;alt=media\" alt=\"\">\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2976720%2F4b52854b32471e8c68297ba46fd36d36%2FQQ20200728100754.jpg?generation=1595902096286583&amp;alt=media\" alt=\"\"></p>\n\n<p>I am using Faster RCNN with pseudo labeling\nMy notebook output submission file is correct when submitted,\nHowever, it didn't work in the competition submission.\nAnybody has some idea?</p>",
  "messages": [
    {
      "id": "948480",
      "postDate": "07/28/2020 02:11:04",
      "content": "<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2976720%2F4b93f043cd4aa12d3b37ab35a3ae8a9c%2FQQ20200728100726.jpg?generation=1595902096112245&amp;alt=media\" alt=\"\">\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2976720%2F4b52854b32471e8c68297ba46fd36d36%2FQQ20200728100754.jpg?generation=1595902096286583&amp;alt=media\" alt=\"\"></p>\n\n<p>I am using Faster RCNN with pseudo labeling\nMy notebook output submission file is correct when submitted,\nHowever, it didn't work in the competition submission.\nAnybody has some idea?</p>",
      "rawMarkdown": "![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2976720%2F4b93f043cd4aa12d3b37ab35a3ae8a9c%2FQQ20200728100726.jpg?generation=1595902096112245&amp;alt=media)\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2976720%2F4b52854b32471e8c68297ba46fd36d36%2FQQ20200728100754.jpg?generation=1595902096286583&amp;alt=media)\n\nI am using Faster RCNN with pseudo labeling\nMy notebook output submission file is correct when submitted,\nHowever, it didn't work in the competition submission.\nAnybody has some idea?",
      "votes": null
    },
    {
      "id": "950477",
      "postDate": "07/29/2020 12:23:52",
      "content": "<p>Do you solve this problem?Although I read some discussion about this error,I still don't konw what can  I do to repair it.If you find a way to slove it,please tell me,thanks.</p>",
      "rawMarkdown": "Do you solve this problem?Although I read some discussion about this error,I still don't konw what can  I do to repair it.If you find a way to slove it,please tell me,thanks.",
      "votes": null
    },
    {
      "id": "951300",
      "postDate": "07/30/2020 03:47:43",
      "content": "<p>Most probably while testing on the private data, there are some images where there will be no predictions. For those images you will have to have the image name (<code>image_id</code>) in the CSV file along with empty output column. That is the <code>PredictionString</code> column has to be empty. If you are not checking for empty outputs during inference, then it will not generate a CSV file. Because for the 11 test images that we are given, all the images have outputs. But it is not so for private test set. I hope this helps.</p>",
      "rawMarkdown": "Most probably while testing on the private data, there are some images where there will be no predictions. For those images you will have to have the image name (`image_id`) in the CSV file along with empty output column. That is the `PredictionString` column has to be empty. If you are not checking for empty outputs during inference, then it will not generate a CSV file. Because for the 11 test images that we are given, all the images have outputs. But it is not so for private test set. I hope this helps.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 950477,
      "author_name": "fengjiedong",
      "author_url": "",
      "post_date": "07/29/2020 12:23:52",
      "content": "<p>Do you solve this problem?Although I read some discussion about this error,I still don't konw what can  I do to repair it.If you find a way to slove it,please tell me,thanks.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 951300,
      "author_name": "sovitrath",
      "author_url": "",
      "post_date": "07/30/2020 03:47:43",
      "content": "<p>Most probably while testing on the private data, there are some images where there will be no predictions. For those images you will have to have the image name (<code>image_id</code>) in the CSV file along with empty output column. That is the <code>PredictionString</code> column has to be empty. If you are not checking for empty outputs during inference, then it will not generate a CSV file. Because for the 11 test images that we are given, all the images have outputs. But it is not so for private test set. I hope this helps.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "948480": "![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2976720%2F4b93f043cd4aa12d3b37ab35a3ae8a9c%2FQQ20200728100726.jpg?generation=1595902096112245&amp;alt=media)\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2976720%2F4b52854b32471e8c68297ba46fd36d36%2FQQ20200728100754.jpg?generation=1595902096286583&amp;alt=media)\n\nI am using Faster RCNN with pseudo labeling\nMy notebook output submission file is correct when submitted,\nHowever, it didn't work in the competition submission.\nAnybody has some idea?",
    "950477": "Do you solve this problem?Although I read some discussion about this error,I still don't konw what can  I do to repair it.If you find a way to slove it,please tell me,thanks.",
    "951300": "Most probably while testing on the private data, there are some images where there will be no predictions. For those images you will have to have the image name (`image_id`) in the CSV file along with empty output column. That is the `PredictionString` column has to be empty. If you are not checking for empty outputs during inference, then it will not generate a CSV file. Because for the 11 test images that we are given, all the images have outputs. But it is not so for private test set. I hope this helps."
  },
  "source": "meta"
}