Hello all! With just over 2 weeks left in this competition, we would like to encourage you to make your submissions for the Judges Prize. Teams will present their submissions to a panel of experts and be judged on their methodology (Scientific Prize, $15,000), innovation and applications (Innovation Prize, $10,000), and diversity and presentation (Diversity Prize, $3,000). The deadline for submissions is May 10th. Details on judging criteria and the submission form can be found here: https://www.kaggle.com/c/hubmap-kidney-segmentation/overview/judges-prize. Get to know our panel of judges via the Organizers & Sponsors page (https://www.kaggle.com/c/hubmap-kidney-segmentation/overview/organizers-and-sponsors). As always, good luck, and we look forward to seeing your contributions!
\nPS - Details on the Judges Prize presentations and competition awards ceremony will be announced soon!
", "messages": [ { "id": 1280983, "postDate": "2021-04-22T14:48:22.700Z", "content": "Hello all! With just over 2 weeks left in this competition, we would like to encourage you to make your submissions for the Judges Prize. Teams will present their submissions to a panel of experts and be judged on their methodology (Scientific Prize, $15,000), innovation and applications (Innovation Prize, $10,000), and diversity and presentation (Diversity Prize, $3,000). The deadline for submissions is May 10th. Details on judging criteria and the submission form can be found here: https://www.kaggle.com/c/hubmap-kidney-segmentation/overview/judges-prize. Get to know our panel of judges via the Organizers & Sponsors page (https://www.kaggle.com/c/hubmap-kidney-segmentation/overview/organizers-and-sponsors). As always, good luck, and we look forward to seeing your contributions!
\nPS - Details on the Judges Prize presentations and competition awards ceremony will be announced soon!
", "rawMarkdown": "Hello all! With just over 2 weeks left in this competition, we would like to encourage you to make your submissions for the **Judges Prize**. Teams will present their submissions to a panel of experts and be judged on their methodology (Scientific Prize, $15,000), innovation and applications (Innovation Prize, $10,000), and diversity and presentation (Diversity Prize, $3,000). The deadline for submissions is **May 10th**. Details on judging criteria and the submission form can be found here: https://www.kaggle.com/c/hubmap-kidney-segmentation/overview/judges-prize. Get to know our panel of judges via the Organizers & Sponsors page (https://www.kaggle.com/c/hubmap-kidney-segmentation/overview/organizers-and-sponsors). As always, good luck, and we look forward to seeing your contributions!\n\nPS - Details on the Judges Prize presentations and competition awards ceremony will be announced soon!", "votes": 9 }, { "id": 1281063, "postDate": "2021-04-22T15:52:47.263Z", "content": "Hi @leahscherschel !
\nIs anyone looking into annotation issues as the one I summarized here ?
\nIt seems tough for anyone to make any submission for the Judges prize while such questions are still unanswered ^^
\nMore generally, I'm really concerned with the quality of the labels in the private test set. If the level of noise is similar to the one in the training data (and public test), then this competition is meaningless.
", "rawMarkdown": "Hi @leahscherschel !\n\nIs anyone looking into annotation issues as the one I summarized [here](https://www.kaggle.com/c/hubmap-kidney-segmentation/discussion/228993#1272538) ?\n\nIt seems tough for anyone to make any submission for the Judges prize while such questions are still unanswered ^^\n\n\nMore generally, I'm really concerned with the quality of the labels in the private test set. If the level of noise is similar to the one in the training data (and public test), then this competition is meaningless. ", "votes": 6, "replies": [ { "id": 1281233, "postDate": "2021-04-22T18:26:29.690Z", "content": "I can't agree more with @theoviel. Hope some key questions have an answer from hosts, specially that ones related with label quality in train and test(Public/Private).
", "rawMarkdown": "I can't agree more with @theoviel. Hope some key questions have an answer from hosts, specially that ones related with label quality in train and test(Public/Private). ", "votes": 4 }, { "id": 1281244, "postDate": "2021-04-22T18:34:54.197Z", "content": "Hi Theo,
\nAll images, public and private alike, went through the same annotation pipeline using the same criteria. Therefore, the annotations should be consistent. Since HuBMAP's focus is on creating a reference atlas of healthy human tissue, globally sclerotic glomeruli were excluded. However, level of injury is on a continuous scale, so some annotated glomeruli are not perfectly healthy. I believe this includes the fibrous crescent glomeruli you referred to above.
\nBest,
\nLeah
", "rawMarkdown": "Hi Theo,\n\nAll images, public and private alike, went through the same annotation pipeline using the same criteria. Therefore, the annotations should be consistent. Since HuBMAP's focus is on creating a reference atlas of healthy human tissue, globally sclerotic glomeruli were excluded. However, level of injury is on a continuous scale, so some annotated glomeruli are not perfectly healthy. I believe this includes the fibrous crescent glomeruli you referred to above.\n\nBest,\n\nLeah", "votes": 6 }, { "id": 1283590, "postDate": "2021-04-25T04:37:09.143Z", "content": "@leahscherschel does it means that its expected to have some more Slides in private set as well with unannotted fibrous crescent glomeruli
", "rawMarkdown": "@leahscherschel does it means that its expected to have some more Slides in private set as well with unannotted fibrous crescent glomeruli ", "votes": 2 }, { "id": 1284785, "postDate": "2021-04-26T08:59:47.483Z", "rawMarkdown": "", "votes": 3, "isDeleted": true }, { "id": 1284908, "postDate": "2021-04-26T11:53:06.863Z", "content": "ask +1. 10 characters
", "rawMarkdown": "ask +1. 10 characters " }, { "id": 1286320, "postDate": "2021-04-27T20:02:26.097Z", "content": "@leahscherschel I agree that we should have a definite answer for the fibrous glomeruli. Some dark glomeruli are annotated in d44c759a, but the same kind of dark glomeruli is not annotated in other images like 57512b7f1 in test set, as I discussed here https://www.kaggle.com/c/hubmap-kidney-segmentation/discussion/234384. The annotations should be consistent.
\nCould you: 1. check the images and confirm the inconsistency of the fibrous glomeruli mentioned above? 2. give us a clear answer whether or not that kind of fibrous glomeruli (that is, dark pink glomeruli with few dark dots inside) should be included? If yes, why not 57512b7f1 doesn't have those annotations?
", "rawMarkdown": "@leahscherschel I agree that we should have a definite answer for the fibrous glomeruli. Some dark glomeruli are annotated in d44c759a, but the same kind of dark glomeruli is not annotated in other images like 57512b7f1 in test set, as I discussed here https://www.kaggle.com/c/hubmap-kidney-segmentation/discussion/234384. The annotations should be consistent. \n\nCould you: 1. check the images and confirm the inconsistency of the fibrous glomeruli mentioned above? 2. give us a clear answer whether or not that kind of fibrous glomeruli (that is, dark pink glomeruli with few dark dots inside) should be included? If yes, why not 57512b7f1 doesn't have those annotations? " }, { "id": 1292237, "postDate": "2021-05-03T19:18:36.387Z", "content": "@leahscherschel All images, public and private alike does thsi also include Train data provided ?
\nplease let us know this
@leahscherschel I've created a kernel about labels inconsistencies (analysis of d488c759a image) that made some of us go to the wrong direction. https://www.kaggle.com/mpware/handlabels-prize
Hi @leahscherschel,
\nDoes the notebook need to be public or just remain private? Thanks.
", "rawMarkdown": "Hi @leahscherschel,\n\nDoes the notebook need to be public or just remain private? Thanks.\n\n", "replies": [ { "id": 1281338, "postDate": "2021-04-22T20:08:34.047Z", "content": "All notebooks submitted must be made public on or before the submission deadline to be eligible.
", "rawMarkdown": "All notebooks submitted must be made public on or before the submission deadline to be eligible." }, { "id": 1284791, "postDate": "2021-04-26T09:09:40.173Z", "content": "Hi @leahscherschel,
\nI am a little confused. Do the notebooks submitted but not for the Judges prize need to be public? Thanks.
Sorry for the confusion. All notebooks submitted for the Judges Prize must be made public on or before the submission deadline to be eligible.
", "rawMarkdown": "Sorry for the confusion. All notebooks submitted *for the Judges Prize* must be made public on or before the submission deadline to be eligible." } ] } ], "comments": [ { "id": 1281063, "author_name": "Theo Viel", "author_url": "", "post_date": "2021-04-22T15:52:47.263000", "content": "Hi @leahscherschel !
\nIs anyone looking into annotation issues as the one I summarized here ?
\nIt seems tough for anyone to make any submission for the Judges prize while such questions are still unanswered ^^
\nMore generally, I'm really concerned with the quality of the labels in the private test set. If the level of noise is similar to the one in the training data (and public test), then this competition is meaningless.
", "votes": 6, "replies": [ { "id": 1281233, "author_name": "Giba", "author_url": "", "post_date": "2021-04-22T18:26:29.690000", "content": "I can't agree more with @theoviel. Hope some key questions have an answer from hosts, specially that ones related with label quality in train and test(Public/Private).
", "votes": 4, "replies": [] }, { "id": 1281244, "author_name": "Leah Godwin", "author_url": "", "post_date": "2021-04-22T18:34:54.197000", "content": "Hi Theo,
\nAll images, public and private alike, went through the same annotation pipeline using the same criteria. Therefore, the annotations should be consistent. Since HuBMAP's focus is on creating a reference atlas of healthy human tissue, globally sclerotic glomeruli were excluded. However, level of injury is on a continuous scale, so some annotated glomeruli are not perfectly healthy. I believe this includes the fibrous crescent glomeruli you referred to above.
\nBest,
\nLeah
", "votes": 6, "replies": [] }, { "id": 1283590, "author_name": "Jaideep", "author_url": "", "post_date": "2021-04-25T04:37:09.143000", "content": "@leahscherschel does it means that its expected to have some more Slides in private set as well with unannotted fibrous crescent glomeruli
", "votes": 2, "replies": [] }, { "id": 1284785, "author_name": "", "author_url": "", "post_date": "2021-04-26T08:59:47.483000", "content": "", "votes": 3, "replies": [] }, { "id": 1284908, "author_name": "朴大福", "author_url": "", "post_date": "2021-04-26T11:53:06.863000", "content": "ask +1. 10 characters
", "votes": 0, "replies": [] }, { "id": 1286320, "author_name": "Swikwislkdjc", "author_url": "", "post_date": "2021-04-27T20:02:26.097000", "content": "@leahscherschel I agree that we should have a definite answer for the fibrous glomeruli. Some dark glomeruli are annotated in d44c759a, but the same kind of dark glomeruli is not annotated in other images like 57512b7f1 in test set, as I discussed here https://www.kaggle.com/c/hubmap-kidney-segmentation/discussion/234384. The annotations should be consistent.
\nCould you: 1. check the images and confirm the inconsistency of the fibrous glomeruli mentioned above? 2. give us a clear answer whether or not that kind of fibrous glomeruli (that is, dark pink glomeruli with few dark dots inside) should be included? If yes, why not 57512b7f1 doesn't have those annotations?
", "votes": 0, "replies": [] }, { "id": 1292237, "author_name": "Jaideep", "author_url": "", "post_date": "2021-05-03T19:18:36.387000", "content": "@leahscherschel All images, public and private alike does thsi also include Train data provided ?
\nplease let us know this
@leahscherschel I've created a kernel about labels inconsistencies (analysis of d488c759a image) that made some of us go to the wrong direction. https://www.kaggle.com/mpware/handlabels-prize
Hi @leahscherschel,
\nDoes the notebook need to be public or just remain private? Thanks.
", "votes": 0, "replies": [ { "id": 1281338, "author_name": "Leah Godwin", "author_url": "", "post_date": "2021-04-22T20:08:34.047000", "content": "All notebooks submitted must be made public on or before the submission deadline to be eligible.
", "votes": 0, "replies": [] }, { "id": 1284791, "author_name": "HayzXiao", "author_url": "", "post_date": "2021-04-26T09:09:40.173000", "content": "Hi @leahscherschel,
\nI am a little confused. Do the notebooks submitted but not for the Judges prize need to be public? Thanks.
Sorry for the confusion. All notebooks submitted for the Judges Prize must be made public on or before the submission deadline to be eligible.
", "votes": 0, "replies": [] } ] } ], "raw_markdown_by_id": { "1280983": "Hello all! With just over 2 weeks left in this competition, we would like to encourage you to make your submissions for the **Judges Prize**. Teams will present their submissions to a panel of experts and be judged on their methodology (Scientific Prize, $15,000), innovation and applications (Innovation Prize, $10,000), and diversity and presentation (Diversity Prize, $3,000). The deadline for submissions is **May 10th**. Details on judging criteria and the submission form can be found here: https://www.kaggle.com/c/hubmap-kidney-segmentation/overview/judges-prize. Get to know our panel of judges via the Organizers & Sponsors page (https://www.kaggle.com/c/hubmap-kidney-segmentation/overview/organizers-and-sponsors). As always, good luck, and we look forward to seeing your contributions!\n\nPS - Details on the Judges Prize presentations and competition awards ceremony will be announced soon!", "1281063": "Hi @leahscherschel !\n\nIs anyone looking into annotation issues as the one I summarized [here](https://www.kaggle.com/c/hubmap-kidney-segmentation/discussion/228993#1272538) ?\n\nIt seems tough for anyone to make any submission for the Judges prize while such questions are still unanswered ^^\n\n\nMore generally, I'm really concerned with the quality of the labels in the private test set. If the level of noise is similar to the one in the training data (and public test), then this competition is meaningless. ", "1281334": "Hi @leahscherschel,\n\nDoes the notebook need to be public or just remain private? Thanks.\n\n" } }