At least just tell us that you've officially stated that there is a problem?
", "messages": [ { "id": 1330964, "postDate": "2021-06-01T07:38:10.347Z", "content": "At least just tell us that you've officially stated that there is a problem?
", "rawMarkdown": "At least just tell us that you've officially stated that there is a problem?", "votes": 13 }, { "id": 1331098, "postDate": "2021-06-01T09:15:48.173Z", "content": "Which bug? (I have not read the forum yet, just looking to see if I'll join next).
\nCan you point to some context?
", "rawMarkdown": "Which bug? (I have not read the forum yet, just looking to see if I'll join next).\n\nCan you point to some context?\n", "votes": 1, "replies": [ { "id": 1331135, "postDate": "2021-06-01T09:47:47.687Z", "content": "The bug is concerning the competition metric. The following discussion contains the issue.
\nhttps://www.kaggle.com/c/siim-covid19-detection/discussion/241291
\nhttps://www.kaggle.com/c/siim-covid19-detection/discussion/241291#1326396
In this competition, there are an OD part and a classification part that are independent. The problem is that the LB doesn't take into account the OD part.
", "rawMarkdown": "In this competition, there are an OD part and a classification part that are independent. The problem is that the LB doesn't take into account the OD part.", "votes": 2 }, { "id": 1331482, "postDate": "2021-06-01T13:52:00.170Z", "content": "I quickly glnced at this, but how can you be usre given LB only shows one number? Is it because your LB score does not change when you change image level predictions?
", "rawMarkdown": "I quickly glnced at this, but how can you be usre given LB only shows one number? Is it because your LB score does not change when you change image level predictions?", "votes": 1 }, { "id": 1331490, "postDate": "2021-06-01T13:55:24.367Z", "content": "I've sent my submissions with and without the boxes (the OD part) and nothing has changed in LB score. Many have reported the same issue as well.
", "rawMarkdown": "I've sent my submissions with and without the boxes (the OD part) and nothing has changed in LB score. Many have reported the same issue as well.", "votes": 2 }, { "id": 1331552, "postDate": "2021-06-01T14:34:05.147Z", "content": "Thanks. I'll wait before joining then.
", "rawMarkdown": "Thanks. I'll wait before joining then.", "votes": 3 }, { "id": 1331557, "postDate": "2021-06-01T14:38:05.730Z", "content": "There is still the classification part that seems to work correctly though!
", "rawMarkdown": "There is still the classification part that seems to work correctly though!" }, { "id": 1331710, "postDate": "2021-06-01T16:44:57.777Z", "content": "lol and I was already wondering why public kernels only fit efnets
", "rawMarkdown": "lol and I was already wondering why public kernels only fit efnets", "votes": 7 }, { "id": 1331741, "postDate": "2021-06-01T17:20:16.290Z", "content": "organizers rushed the competition… there is a lot of mess with the data also!
", "rawMarkdown": "organizers rushed the competition... there is a lot of mess with the data also!", "votes": 12 }, { "id": 1332035, "postDate": "2021-06-01T21:28:17.337Z", "content": "that's a shame… and noone fixing it?
", "rawMarkdown": "that's a shame... and noone fixing it?" }, { "id": 1332132, "postDate": "2021-06-01T23:55:51.197Z", "content": "They said they will fix everything by next week
\nhttps://www.kaggle.com/c/siim-covid19-detection/discussion/241291#1331802
What about outliars that r pointed out by public kernels
\nie. discrepancy between Image level labels and study level . I.e no bbx given for Non Negative Images. @cpmpml @debarshichanda @morizin @philippsinger etc
I think the hosts are working towards correcting it
\nhttps://www.kaggle.com/c/siim-covid19-detection/discussion/240250#1322940
\nhttps://www.kaggle.com/c/siim-covid19-detection/discussion/240250#1335165
actually I took part in a similar mini competition of object detection and classification, and overall scene classification. Their score system was messed up at beginning too. this should be two tasks, separately or with some weights to score.
", "rawMarkdown": "actually I took part in a similar mini competition of object detection and classification, and overall scene classification. Their score system was messed up at beginning too. this should be two tasks, separately or with some weights to score." } ], "comments": [ { "id": 1331098, "author_name": "CPMP", "author_url": "", "post_date": "2021-06-01T09:15:48.173000", "content": "Which bug? (I have not read the forum yet, just looking to see if I'll join next).
\nCan you point to some context?
", "votes": 1, "replies": [ { "id": 1331135, "author_name": "Mohammad Zunaed", "author_url": "", "post_date": "2021-06-01T09:47:47.687000", "content": "The bug is concerning the competition metric. The following discussion contains the issue.
\nhttps://www.kaggle.com/c/siim-covid19-detection/discussion/241291
\nhttps://www.kaggle.com/c/siim-covid19-detection/discussion/241291#1326396
In this competition, there are an OD part and a classification part that are independent. The problem is that the LB doesn't take into account the OD part.
", "votes": 2, "replies": [] }, { "id": 1331482, "author_name": "CPMP", "author_url": "", "post_date": "2021-06-01T13:52:00.170000", "content": "I quickly glnced at this, but how can you be usre given LB only shows one number? Is it because your LB score does not change when you change image level predictions?
", "votes": 1, "replies": [] }, { "id": 1331490, "author_name": "Joseph AMIGO", "author_url": "", "post_date": "2021-06-01T13:55:24.367000", "content": "I've sent my submissions with and without the boxes (the OD part) and nothing has changed in LB score. Many have reported the same issue as well.
", "votes": 2, "replies": [] }, { "id": 1331552, "author_name": "CPMP", "author_url": "", "post_date": "2021-06-01T14:34:05.147000", "content": "Thanks. I'll wait before joining then.
", "votes": 3, "replies": [] }, { "id": 1331557, "author_name": "Joseph AMIGO", "author_url": "", "post_date": "2021-06-01T14:38:05.730000", "content": "There is still the classification part that seems to work correctly though!
", "votes": 0, "replies": [] }, { "id": 1331710, "author_name": "Psi", "author_url": "", "post_date": "2021-06-01T16:44:57.777000", "content": "lol and I was already wondering why public kernels only fit efnets
", "votes": 7, "replies": [] }, { "id": 1331741, "author_name": "raddar", "author_url": "", "post_date": "2021-06-01T17:20:16.290000", "content": "organizers rushed the competition… there is a lot of mess with the data also!
", "votes": 12, "replies": [] }, { "id": 1332035, "author_name": "Psi", "author_url": "", "post_date": "2021-06-01T21:28:17.337000", "content": "that's a shame… and noone fixing it?
", "votes": 0, "replies": [] }, { "id": 1332132, "author_name": "Debarshi Chanda", "author_url": "", "post_date": "2021-06-01T23:55:51.197000", "content": "They said they will fix everything by next week
\nhttps://www.kaggle.com/c/siim-covid19-detection/discussion/241291#1331802
What about outliars that r pointed out by public kernels
\nie. discrepancy between Image level labels and study level . I.e no bbx given for Non Negative Images. @cpmpml @debarshichanda @morizin @philippsinger etc
I think the hosts are working towards correcting it
\nhttps://www.kaggle.com/c/siim-covid19-detection/discussion/240250#1322940
\nhttps://www.kaggle.com/c/siim-covid19-detection/discussion/240250#1335165
actually I took part in a similar mini competition of object detection and classification, and overall scene classification. Their score system was messed up at beginning too. this should be two tasks, separately or with some weights to score.
", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "1330964": "At least just tell us that you've officially stated that there is a problem?", "1331098": "Which bug? (I have not read the forum yet, just looking to see if I'll join next).\n\nCan you point to some context?\n", "1335756": "actually I took part in a similar mini competition of object detection and classification, and overall scene classification. Their score system was messed up at beginning too. this should be two tasks, separately or with some weights to score." } }