![\"img1\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F720378%2Fde2df20cb131ae96b1f4a2a546a33de2%2FSnipaste_2024-09-19_23-25-52.png?generation=1726759583650120&alt=media\")
\n
![\"img2\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F720378%2Fb0638b8f9d1fa7483b75b0410a37c301%2FSnipaste_2024-09-19_23-26-04.png?generation=1726759603439760&alt=media\")
While analyzing the T2 data related to Canal Stenosis, I found annotation errors at some key points. If these erroneous data are used to train the model, it may interfere with the model's performance. You can choose to delete these data or correct them.You can check this kernel to see how I identified these problematic key points.
\nOne more thing, I am sharing this because I joined the competition too late and I am a beginner in the field of object detection. Therefore, if this discovery is useful to you, I will be very happy.
\n
\n
While analyzing the T2 data related to Canal Stenosis, I found annotation errors at some key points. If these erroneous data are used to train the model, it may interfere with the model's performance. You can choose to delete these data or correct them.You can check this kernel to see how I identified these problematic key points.
\nOne more thing, I am sharing this because I joined the competition too late and I am a beginner in the field of object detection. Therefore, if this discovery is useful to you, I will be very happy.
\n
\n
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F6882624%2F01363ec3655d163026114624797663c7%2FScreenshot%20from%202024-09-19%2017-34-27.png?generation=1726760310786493&alt=media\")
Yes, there are 13 values with x and y = 5.
Hi @mnk812, it been the known problem for 3 months already. There are many inconsistencies with the data. You can take your time and thoroughly go through discussions and your own validation.
\n", "rawMarkdown": "Hi @mnk812, it been the known problem for 3 months already. There are many inconsistencies with the data. You can take your time and thoroughly go through discussions and your own validation.\n\n[check this out](https://www.kaggle.com/competitions/rsna-2024-lumbar-spine-degenerative-classification/discussion/512048)", "votes": null }, { "id": "2993619", "postDate": "09/20/2024 02:22:09", "content": "Thank you, I didn't notice this discussion.๐
", "rawMarkdown": "Thank you, I didn't notice this discussion.๐ ", "votes": null }, { "id": "2997133", "postDate": "09/24/2024 09:16:51", "content": "This problem seems to have been solved.
", "rawMarkdown": "This problem seems to have been solved.", "votes": null } ], "comments": [ { "id": 2993266, "author_name": "jankowalski2000", "author_url": "", "post_date": "09/19/2024 15:37:32", "content": " Yes, there are 13 values with x and y = 5.
Hi @mnk812, it been the known problem for 3 months already. There are many inconsistencies with the data. You can take your time and thoroughly go through discussions and your own validation.
\n", "votes": null, "replies": [ { "id": 2993619, "author_name": "mnk812", "author_url": "", "post_date": "09/20/2024 02:22:09", "content": "Thank you, I didn't notice this discussion.๐
", "votes": null, "replies": [] } ] }, { "id": 2997133, "author_name": "", "author_url": "", "post_date": "09/24/2024 09:16:51", "content": "This problem seems to have been solved.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "2993261": "While analyzing the T2 data related to Canal Stenosis, I found annotation errors at some key points. If these erroneous data are used to train the model, it may interfere with the model's performance. You can choose to delete these data or correct them.You can check [this kernel](https://www.kaggle.com/code/mnk812/20240919-make-t2-img/notebook) to see how I identified these problematic key points.\n\nOne more thing, I am sharing this because I joined the competition too late and I am a beginner in the field of object detection. Therefore, if this discovery is useful to you, I will be very happy.\n\n\n", "2993266": " Yes, there are 13 values with x and y = 5.", "2993300": "Hi @mnk812, it been the known problem for 3 months already. There are many inconsistencies with the data. You can take your time and thoroughly go through discussions and your own validation.\n\n[check this out](https://www.kaggle.com/competitions/rsna-2024-lumbar-spine-degenerative-classification/discussion/512048)", "2993619": "Thank you, I didn't notice this discussion.๐ ", "2997133": "This problem seems to have been solved." }, "source": "meta" }