\n
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F6400548%2F2dbc4acb974adc0a260f589487a4dbd2%2F1.2.826.0.1.3680043.8.498.10017151185038598671064029373595425922.jpg?generation=1610548085014451&alt=media\")
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F6400548%2Fe7f6c9b8b5db6534b3ff881a7d40ddc2%2F1.2.826.0.1.3680043.8.498.10128199840925023408402703824969146572.jpg?generation=1610548575851076&alt=media\")
\nThanks for discussing!
Our model has low scores on the CVC class, so we visualize the tube according the file{'train_annotations'}. we get the image as follow,but we think the tube's position maybe not right in the CVC_borderline. Does anybody have the same idea as us? How do you deal this incorrect image? Delete or ignore? And your method ?
\n
\nThanks for discussing!
Our model has low scores on the CVC class, so we visualize the tube according the file{'train_annotations'}. we get the image as follow,but we think the tube's position maybe not right in the CVC_borderline. Does anybody have the same idea as us? How do you deal this incorrect image? Delete or ignore? And your method ?
\n
\nThanks for discussing!
This image and borderline class is completely fine by the borderline definitinon in this competition
", "rawMarkdown": "This image and borderline class is completely fine by the borderline definitinon in this competition", "votes": null }, { "id": "1152572", "postDate": "01/14/2021 09:24:02", "content": "I wonder if it was because one end of the catheter was inserted into the vein in the left arm (the red square area marked in my picture below), so the image was divided into borderline.I'm really curious, because the two above image are very different from those in other normal image. The normal catheter ports are located in the veins of the heart.
", "rawMarkdown": "I wonder if it was because one end of the catheter was inserted into the vein in the left arm (the red square area marked in my picture below), so the image was divided into borderline.I'm really curious, because the two above image are very different from those in other normal image. The normal catheter ports are located in the veins of the heart.[](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F6400548%2Ff57a8476598f67ca1dad6d0d8fed3f31%2FBlausen_0181_Catheter_CentralVenousAccessDevice_NonTunneled.png?generation=1610615274235141&alt=media)", "votes": null }, { "id": "1152573", "postDate": "01/14/2021 09:27:29", "content": "![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F6400548%2F8f6aa7d4603d04a9bd897a992b33587b%2FBlausen_0181_Catheter_CentralVenousAccessDevice_NonTunneled.png?generation=1610616446606319&alt=media\")
Hi,I also finid this problem,in my predicted result,I find the 'CVC-borderline' class auc score is the lowest.How do you solve this problem?Thanks!
", "rawMarkdown": "Hi,I also finid this problem,in my predicted result,I find the 'CVC-borderline' class auc score is the lowest.How do you solve this problem?Thanks!", "votes": null } ], "comments": [ { "id": 1151864, "author_name": "raddar", "author_url": "", "post_date": "01/13/2021 16:08:34", "content": "This image and borderline class is completely fine by the borderline definitinon in this competition
", "votes": null, "replies": [ { "id": 1152572, "author_name": "baixueya", "author_url": "", "post_date": "01/14/2021 09:24:02", "content": "I wonder if it was because one end of the catheter was inserted into the vein in the left arm (the red square area marked in my picture below), so the image was divided into borderline.I'm really curious, because the two above image are very different from those in other normal image. The normal catheter ports are located in the veins of the heart.
", "votes": null, "replies": [] }, { "id": 1152573, "author_name": "baixueya", "author_url": "", "post_date": "01/14/2021 09:27:29", "content": "
Hi,I also finid this problem,in my predicted result,I find the 'CVC-borderline' class auc score is the lowest.How do you solve this problem?Thanks!
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1151759": "Our model has low scores on the CVC class, so we visualize the tube according the file{'train_annotations'}. we get the image as follow,but we think the tube's position maybe not right in the CVC_borderline. Does anybody have the same idea as us? How do you deal this incorrect image? Delete or ignore? And your method ?\n\nThanks for discussing!", "1151864": "This image and borderline class is completely fine by the borderline definitinon in this competition", "1152572": "I wonder if it was because one end of the catheter was inserted into the vein in the left arm (the red square area marked in my picture below), so the image was divided into borderline.I'm really curious, because the two above image are very different from those in other normal image. The normal catheter ports are located in the veins of the heart.[](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F6400548%2Ff57a8476598f67ca1dad6d0d8fed3f31%2FBlausen_0181_Catheter_CentralVenousAccessDevice_NonTunneled.png?generation=1610615274235141&alt=media)", "1152573": "", "1167570": "Hi,I also finid this problem,in my predicted result,I find the 'CVC-borderline' class auc score is the lowest.How do you solve this problem?Thanks!" }, "source": "meta" }