{ "id": 492891, "title": "Inconsistency between description and data", "url": "/competitions/image-matching-challenge-2024/discussion/492891", "author_name": "", "post_date": "2024-04-11T10:16:47.441778100Z", "votes": null, "comment_count": 4, "views": 0, "content": "

I have noticed that the description of competition data suggest that there is subdirectory scene in the dataset (all represented as /* in quote below)

\n
\n

[train/test]/*/*/images

\n
\n

However the sample_submission.csv and data explorer all suggest that there are is only subdirectory dataset. In 2023 competition, there are many scenes in the same datasets, necessitating the use of two-level categorization, but there is only one scene in every train and test dataset this year. I wonder if there is a mistake in the description.

\n

EDIT: the sample_submission.csv looks quite confusing
\n

", "messages": [ { "id": "2746559", "postDate": "04/11/2024 10:16:47", "content": "

I have noticed that the description of competition data suggest that there is subdirectory scene in the dataset (all represented as /* in quote below)

\n
\n

[train/test]/*/*/images

\n
\n

However the sample_submission.csv and data explorer all suggest that there are is only subdirectory dataset. In 2023 competition, there are many scenes in the same datasets, necessitating the use of two-level categorization, but there is only one scene in every train and test dataset this year. I wonder if there is a mistake in the description.

\n

EDIT: the sample_submission.csv looks quite confusing
\n

", "rawMarkdown": "I have noticed that the description of competition data suggest that there is subdirectory `scene` in the `dataset` (all represented as `/*` in quote below)\n\n>[train/test]/\\*/\\*/images \n\nHowever the `sample_submission.csv` and data explorer all suggest that there are is only subdirectory dataset. In 2023 competition, there are many scenes in the same datasets, necessitating the use of two-level categorization, but there is only one scene in every train and test dataset this year. I wonder if there is a mistake in the description. \n\nEDIT: the sample_submission.csv looks quite confusing\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F13889710%2Fa78ccf0577f71cfbe5b016431f2d68cb%2FScreenshot%202024-04-11%20at%206.25.19PM.png?generation=1712831163930748&alt=media)", "votes": null }, { "id": "2747851", "postDate": "04/12/2024 05:33:03", "content": "

@oldufo do you mind taking a look into this?

", "rawMarkdown": "oldufo do you mind taking a look into this?", "votes": null }, { "id": "2748061", "postDate": "04/12/2024 08:20:46", "content": "

@renyiwei thank you, the description is fixed now to [train/test]/*/images

", "rawMarkdown": "renyiwei thank you, the description is fixed now to `[train/test]/*/images`", "votes": null }, { "id": "2748347", "postDate": "04/12/2024 11:58:31", "content": "

@oldufo Actually, it is better to just remove the idea of scene all together. That said, it's a lot of work to revise it as everyone then needs to make adjustment in their notebooks.

", "rawMarkdown": "oldufo Actually, it is better to just remove the idea of `scene` all together. That said, it's a lot of work to revise it as everyone then needs to make adjustment in their notebooks.", "votes": null }, { "id": "2748356", "postDate": "04/12/2024 12:04:33", "content": "

Too late, I believe. Sorry for the inconvenience.

", "rawMarkdown": "Too late, I believe. Sorry for the inconvenience.", "votes": null } ], "comments": [ { "id": 2747851, "author_name": "renyiwei", "author_url": "", "post_date": "04/12/2024 05:33:03", "content": "

@oldufo do you mind taking a look into this?

", "votes": null, "replies": [] }, { "id": 2748061, "author_name": "oldufo", "author_url": "", "post_date": "04/12/2024 08:20:46", "content": "

@renyiwei thank you, the description is fixed now to [train/test]/*/images

", "votes": null, "replies": [ { "id": 2748347, "author_name": "renyiwei", "author_url": "", "post_date": "04/12/2024 11:58:31", "content": "

@oldufo Actually, it is better to just remove the idea of scene all together. That said, it's a lot of work to revise it as everyone then needs to make adjustment in their notebooks.

", "votes": null, "replies": [ { "id": 2748356, "author_name": "oldufo", "author_url": "", "post_date": "04/12/2024 12:04:33", "content": "

Too late, I believe. Sorry for the inconvenience.

", "votes": null, "replies": [] } ] } ] } ], "raw_markdown_by_id": { "2746559": "I have noticed that the description of competition data suggest that there is subdirectory `scene` in the `dataset` (all represented as `/*` in quote below)\n\n>[train/test]/\\*/\\*/images \n\nHowever the `sample_submission.csv` and data explorer all suggest that there are is only subdirectory dataset. In 2023 competition, there are many scenes in the same datasets, necessitating the use of two-level categorization, but there is only one scene in every train and test dataset this year. I wonder if there is a mistake in the description. \n\nEDIT: the sample_submission.csv looks quite confusing\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F13889710%2Fa78ccf0577f71cfbe5b016431f2d68cb%2FScreenshot%202024-04-11%20at%206.25.19PM.png?generation=1712831163930748&alt=media)", "2747851": "oldufo do you mind taking a look into this?", "2748061": "renyiwei thank you, the description is fixed now to `[train/test]/*/images`", "2748347": "oldufo Actually, it is better to just remove the idea of `scene` all together. That said, it's a lot of work to revise it as everyone then needs to make adjustment in their notebooks.", "2748356": "Too late, I believe. Sorry for the inconvenience." }, "source": "meta" }