{ "id": 240080, "title": "External Data Thread", "url": "/competitions/siim-covid19-detection/discussion/240080", "author_name": "Kerem Turgutlu", "post_date": "2021-05-18T13:53:33.518000", "votes": 16, "comment_count": 20, "views": 0, "content": "

Freely & publicly available external data is allowed, including pre-trained models

\n

Feel free to extend it.

\n\n

Bonus: COVID-Net

", "messages": [ { "id": 1313298, "postDate": "2021-05-18T13:53:33.517Z", "content": "

Freely & publicly available external data is allowed, including pre-trained models

\n

Feel free to extend it.

\n\n

Bonus: COVID-Net

", "rawMarkdown": "`Freely & publicly available external data is allowed, including pre-trained models`\n\nFeel free to extend it.\n\n- [MIMIC-CXR] (https://physionet.org/content/mimic-cxr/2.0.0/) (Requires submission for access but its a pretty fast process)\n- [MIMIC-CXR JPG] (https://physionet.org/content/mimic-cxr-jpg/2.0.0/) (Requires submission for access but its a pretty fast process)\n- [ChexPert](https://stanfordmlgroup.github.io/competitions/chexpert/) (Requires submission. There are external providers too including [tf](https://www.tensorflow.org/datasets/catalog/chexpert) and [small kaggle](https://www.kaggle.com/awsaf49/chexpert-dataset))\n- https://github.com/ieee8023/covid-chestxray-dataset\n\nBonus: [COVID-Net] (https://github.com/lindawangg/COVID-Net)", "votes": 16 }, { "id": 1322922, "postDate": "2021-05-25T19:35:49.920Z", "content": "

FYI, COVID-Net data can also be found in this kaggle dataset.

\n

I'm just going to shamelessly advertised for COVID-Net as it came out from my lab. :)

", "rawMarkdown": "FYI, COVID-Net data can also be found in this [kaggle dataset](https://www.kaggle.com/andyczhao/covidx-cxr2).\n\nI'm just going to shamelessly advertised for COVID-Net as it came out from my lab. :)", "votes": 4 }, { "id": 3369036, "postDate": "2025-12-09T17:29:35.687Z", "content": "

i need to use the dataset MIMIC-CXR JPG. pls guide me how to use that. what are the procedures to use

", "rawMarkdown": "i need to use the dataset MIMIC-CXR JPG. pls guide me how to use that. what are the procedures to use" }, { "id": 1386522, "postDate": "2021-07-13T14:29:23.197Z", "content": "

Hi,
\nAre public Kaggle data like

\n", "rawMarkdown": "Hi,\nAre public Kaggle data like\n* https://www.kaggle.com/tawsifurrahman/covid19-radiography-database\n* https://www.kaggle.com/andyczhao/covidx-cxr2\n* https://www.kaggle.com/preetviradiya/covid19-radiography-dataset\n* https://www.kaggle.com/prashant268/chest-xray-covid19-pneumonia\n* https://www.kaggle.com/paultimothymooney/chest-xray-pneumonia\n* https://www.kaggle.com/praveengovi/coronahack-chest-xraydataset\n* https://www.kaggle.com/pcbreviglieri/pneumonia-xray-images\n* https://www.kaggle.com/tolgadincer/labeled-chest-xray-images\n* https://www.kaggle.com/unaissait/curated-chest-xray-image-dataset-for-covid19\n* https://www.kaggle.com/edoardovantaggiato/covid19-xray-two-proposed-databases\n* https://www.kaggle.com/endiqq/largest-covid19-dataset\nalso available?\nThanks!" }, { "id": 1331795, "postDate": "2021-06-01T18:15:28.517Z", "content": "

I'm sorry if I'm wrong, but isn't the mimic data only available to researchers?
\nEven if not, we can't download the data without taking a test. Do we have permission to use the mimic data?

", "rawMarkdown": "I'm sorry if I'm wrong, but isn't the mimic data only available to researchers?\nEven if not, we can't download the data without taking a test. Do we have permission to use the mimic data?", "replies": [ { "id": 1332337, "postDate": "2021-06-02T04:39:05.333Z", "content": "

@juliaelliott
\nI think it should be prohibited, but I didn't understand it when I looked at rule section 7C, so please let me know.

", "rawMarkdown": "@juliaelliott \nI think it should be prohibited, but I didn't understand it when I looked at rule section 7C, so please let me know." }, { "id": 1353233, "postDate": "2021-06-17T02:08:18.967Z", "content": "

I am also waiting from kaggle staff to respond before training any models on MIMIC-CXR. Its a lot of images :D

", "rawMarkdown": "I am also waiting from kaggle staff to respond before training any models on MIMIC-CXR. Its a lot of images :D" }, { "id": 1353244, "postDate": "2021-06-17T02:26:40.400Z", "content": "

I think mimic not allowed. In a competition, it was answered by kaggle staff.

", "rawMarkdown": "I think **mimic** not allowed. In a competition, it was answered by kaggle staff." }, { "id": 1354578, "postDate": "2021-06-17T17:11:39.713Z", "content": "

If a dataset is only available under restriction/permission, it is by definition not publicly and freely available, and would therefore be prohibited.

", "rawMarkdown": "If a dataset is only available under restriction/permission, it is by definition **not publicly and freely available**, and would therefore be prohibited.", "votes": 1 }, { "id": 1354599, "postDate": "2021-06-17T17:42:14.250Z", "content": "

@juliaelliott one confusion. Most of the Open Source datasets including ImageNet have one common data usage policy,

\n
for non-commercial use by the research and education communities. \n
\n

Does it mean it's prohibited?

", "rawMarkdown": "@juliaelliott one confusion. Most of the Open Source datasets including **ImageNet** have one common data usage policy, \n```\nfor non-commercial use by the research and education communities. \n```\nDoes it mean it's prohibited?" }, { "id": 1354604, "postDate": "2021-06-17T17:45:23.510Z", "content": "

Also competition data source RICORD has a similar data usage policy.

", "rawMarkdown": "Also competition data source **RICORD** has a similar data usage policy." }, { "id": 1354655, "postDate": "2021-06-17T18:38:05.670Z", "content": "

it is mentioned by the competition host:
\nhttps://www.kaggle.com/c/siim-covid19-detection/discussion/243804#1341447

\n

\"Both RICORD and BIMCV are public datasets\" …

\n

\"Because the goal in the competition is for Kaggler's to develop the best possible model, and with access to the same data for fairness-sake, it is acceptable for the contestants to use the RICORD data with their annotations and also to use BIMCV data with their annotations.\"

", "rawMarkdown": "it is mentioned by the competition host:\nhttps://www.kaggle.com/c/siim-covid19-detection/discussion/243804#1341447\n\n\"Both RICORD and BIMCV are public datasets\" ...\n\n\"Because the goal in the competition is for Kaggler's to develop the best possible model, and with access to the same data for fairness-sake, it is acceptable for the contestants to use the RICORD data with their annotations and also to use BIMCV data with their annotations.\"\n", "votes": 1 }, { "id": 1354659, "postDate": "2021-06-17T18:42:53.043Z", "content": "

So that means we can not use MIMIC-CXR or ChexPert as they require submission/permission to download the data?

", "rawMarkdown": "So that means we can not use MIMIC-CXR or ChexPert as they require submission/permission to download the data? " }, { "id": 1354667, "postDate": "2021-06-17T18:53:31.343Z", "content": "

@awsaf49

\n

\"Most of the Open Source datasets including ImageNet have one common data usage policy,\"

\n

this is something interesting … conferences like cvpr, iccv, nips also held competitions. There are money awards and participants use ImageNet. In fact, there is ImageNet competition itself.

\n

i would interpret as \"money awards does not mean commercial use\". Although there are money awards , the end product (to SIIM) is for research education purpose.

", "rawMarkdown": "@awsaf49 \n\n\"Most of the Open Source datasets including ImageNet have one common data usage policy,\"\n\nthis is something interesting ... conferences like cvpr, iccv, nips also held competitions. There are money awards and participants use ImageNet. In fact, there is ImageNet competition itself.\n\ni would interpret as \"money awards does not mean commercial use\". Although there are money awards , the end product (to SIIM) is for research education purpose.", "votes": 1 }, { "id": 1354709, "postDate": "2021-06-17T19:28:05.427Z", "content": "

So, after lot of searching this is what I got,

\n", "rawMarkdown": "So, after lot of searching this is what I got,\n\n* **CheXpert**,\n@juliaelliott said [here](https://www.kaggle.com/c/siim-acr-pneumothorax-segmentation/discussion/97770#602991),\n\"chexpert-allowed\"\n\n* **MIMIC**:\nas **mimic** requires a **test** to get access,\n@juliaelliott said [here](https://www.kaggle.com/c/siim-acr-pneumothorax-segmentation/discussion/97770#578263),\n\"mimic-not-allowed\"\n\n* **NIH** or **ChestX-ray14**:\ngot a nice license **CC0** we should be able to do anything with it\n\n* **RICORD**:\ngot **Attribution-NonCommercial 4.0 International (CC BY-NC 4.0)** lincese\n\n* **BIMCV**:\naccording to the paper, `It is freely available for\nresearch, and can also be used for commercial purposes under certain conditions\n`\n\n", "votes": 5 }, { "id": 1354712, "postDate": "2021-06-17T19:30:07.910Z", "content": "

thanks @hengck23, this is what I was looking for

\n
\"money awards does not mean commercial use\"\n
", "rawMarkdown": "thanks @hengck23, this is what I was looking for\n```\n\"money awards does not mean commercial use\"\n```" }, { "id": 1354742, "postDate": "2021-06-17T20:08:24.573Z", "content": "

If a dataset is restricted/gated for access/download and therefore non-public and inaccessible to all for free, this would be prohibited, as it is not publicly and freely available to all participants in the competition.

\n

On the other hand, if a dataset is licensed for a research or academic use, this would still be permitted for use, so long as those who choose to use any such datasets ensure that they are able to comply with the licensing requirements of the solution/code they produce per the competition's rules.

", "rawMarkdown": "If a dataset is restricted/gated for access/download and therefore non-public and inaccessible to all for free, this would be prohibited, as it is not **publicly and freely available to all participants** in the competition.\n\nOn the other hand, if a dataset is *licensed for a research or academic use*, this would still be permitted for use, so long as those who choose to use any such datasets ensure that they are able to comply with the licensing requirements of the *solution/code* they produce per the competition's rules.", "votes": 4 }, { "id": 1380782, "postDate": "2021-07-08T10:01:58.870Z", "content": "

Hello, did you use external data?

", "rawMarkdown": "Hello, did you use external data?" }, { "id": 1451333, "postDate": "2021-08-05T10:16:40.987Z", "content": "

@juliaelliott
\nShall we regard this dataset for usage, its easily downloadable by all
\nRICORD:
\ngot Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) lincese

\n

similarly is licensing for covidxr2 which is mix of multiple datasets some of them are GNu license some are Attribution licenses but easily downloadable by all.

", "rawMarkdown": "@juliaelliott \nShall we regard this dataset for usage, its easily downloadable by all \nRICORD:\ngot Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) lincese\n\nsimilarly is licensing for covidxr2 which is mix of multiple datasets some of them are GNu license some are Attribution licenses but easily downloadable by all." } ] }, { "id": 1328152, "postDate": "2021-05-30T00:53:43.510Z", "rawMarkdown": "", "isDeleted": true }, { "id": 1328147, "postDate": "2021-05-30T00:40:06.173Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 1322922, "author_name": "JunYong Tong", "author_url": "", "post_date": "2021-05-25T19:35:49.920000", "content": "

FYI, COVID-Net data can also be found in this kaggle dataset.

\n

I'm just going to shamelessly advertised for COVID-Net as it came out from my lab. :)

", "votes": 4, "replies": [] }, { "id": 3369036, "author_name": "Sangeetha87", "author_url": "", "post_date": "2025-12-09T17:29:35.687000", "content": "

i need to use the dataset MIMIC-CXR JPG. pls guide me how to use that. what are the procedures to use

", "votes": 0, "replies": [] }, { "id": 1386522, "author_name": "lazyterence", "author_url": "", "post_date": "2021-07-13T14:29:23.197000", "content": "

Hi,
\nAre public Kaggle data like

\n", "votes": 0, "replies": [] }, { "id": 1331795, "author_name": "YujiAriyasu", "author_url": "", "post_date": "2021-06-01T18:15:28.517000", "content": "

I'm sorry if I'm wrong, but isn't the mimic data only available to researchers?
\nEven if not, we can't download the data without taking a test. Do we have permission to use the mimic data?

", "votes": 0, "replies": [ { "id": 1332337, "author_name": "YujiAriyasu", "author_url": "", "post_date": "2021-06-02T04:39:05.333000", "content": "

@juliaelliott
\nI think it should be prohibited, but I didn't understand it when I looked at rule section 7C, so please let me know.

", "votes": 0, "replies": [] }, { "id": 1353233, "author_name": "Kerem Turgutlu", "author_url": "", "post_date": "2021-06-17T02:08:18.967000", "content": "

I am also waiting from kaggle staff to respond before training any models on MIMIC-CXR. Its a lot of images :D

", "votes": 0, "replies": [] }, { "id": 1353244, "author_name": "Awsaf", "author_url": "", "post_date": "2021-06-17T02:26:40.400000", "content": "

I think mimic not allowed. In a competition, it was answered by kaggle staff.

", "votes": 0, "replies": [] }, { "id": 1354578, "author_name": "Julia Elliott", "author_url": "", "post_date": "2021-06-17T17:11:39.713000", "content": "

If a dataset is only available under restriction/permission, it is by definition not publicly and freely available, and would therefore be prohibited.

", "votes": 1, "replies": [] }, { "id": 1354599, "author_name": "Awsaf", "author_url": "", "post_date": "2021-06-17T17:42:14.250000", "content": "

@juliaelliott one confusion. Most of the Open Source datasets including ImageNet have one common data usage policy,

\n
for non-commercial use by the research and education communities. \n
\n

Does it mean it's prohibited?

", "votes": 0, "replies": [] }, { "id": 1354604, "author_name": "Awsaf", "author_url": "", "post_date": "2021-06-17T17:45:23.510000", "content": "

Also competition data source RICORD has a similar data usage policy.

", "votes": 0, "replies": [] }, { "id": 1354655, "author_name": "hengck23", "author_url": "", "post_date": "2021-06-17T18:38:05.670000", "content": "

it is mentioned by the competition host:
\nhttps://www.kaggle.com/c/siim-covid19-detection/discussion/243804#1341447

\n

\"Both RICORD and BIMCV are public datasets\" …

\n

\"Because the goal in the competition is for Kaggler's to develop the best possible model, and with access to the same data for fairness-sake, it is acceptable for the contestants to use the RICORD data with their annotations and also to use BIMCV data with their annotations.\"

", "votes": 1, "replies": [] }, { "id": 1354659, "author_name": "Mohammad Zunaed", "author_url": "", "post_date": "2021-06-17T18:42:53.043000", "content": "

So that means we can not use MIMIC-CXR or ChexPert as they require submission/permission to download the data?

", "votes": 0, "replies": [] }, { "id": 1354667, "author_name": "hengck23", "author_url": "", "post_date": "2021-06-17T18:53:31.343000", "content": "

@awsaf49

\n

\"Most of the Open Source datasets including ImageNet have one common data usage policy,\"

\n

this is something interesting … conferences like cvpr, iccv, nips also held competitions. There are money awards and participants use ImageNet. In fact, there is ImageNet competition itself.

\n

i would interpret as \"money awards does not mean commercial use\". Although there are money awards , the end product (to SIIM) is for research education purpose.

", "votes": 1, "replies": [] }, { "id": 1354709, "author_name": "Awsaf", "author_url": "", "post_date": "2021-06-17T19:28:05.427000", "content": "

So, after lot of searching this is what I got,

\n", "votes": 5, "replies": [] }, { "id": 1354712, "author_name": "Awsaf", "author_url": "", "post_date": "2021-06-17T19:30:07.910000", "content": "

thanks @hengck23, this is what I was looking for

\n
\"money awards does not mean commercial use\"\n
", "votes": 0, "replies": [] }, { "id": 1354742, "author_name": "Julia Elliott", "author_url": "", "post_date": "2021-06-17T20:08:24.573000", "content": "

If a dataset is restricted/gated for access/download and therefore non-public and inaccessible to all for free, this would be prohibited, as it is not publicly and freely available to all participants in the competition.

\n

On the other hand, if a dataset is licensed for a research or academic use, this would still be permitted for use, so long as those who choose to use any such datasets ensure that they are able to comply with the licensing requirements of the solution/code they produce per the competition's rules.

", "votes": 4, "replies": [] }, { "id": 1380782, "author_name": "Zekun", "author_url": "", "post_date": "2021-07-08T10:01:58.870000", "content": "

Hello, did you use external data?

", "votes": 0, "replies": [] }, { "id": 1451333, "author_name": "Jaideep", "author_url": "", "post_date": "2021-08-05T10:16:40.987000", "content": "

@juliaelliott
\nShall we regard this dataset for usage, its easily downloadable by all
\nRICORD:
\ngot Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) lincese

\n

similarly is licensing for covidxr2 which is mix of multiple datasets some of them are GNu license some are Attribution licenses but easily downloadable by all.

", "votes": 0, "replies": [] } ] }, { "id": 1328152, "author_name": "", "author_url": "", "post_date": "2021-05-30T00:53:43.510000", "content": "", "votes": 0, "replies": [] }, { "id": 1328147, "author_name": "", "author_url": "", "post_date": "2021-05-30T00:40:06.173000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "1313298": "`Freely & publicly available external data is allowed, including pre-trained models`\n\nFeel free to extend it.\n\n- [MIMIC-CXR] (https://physionet.org/content/mimic-cxr/2.0.0/) (Requires submission for access but its a pretty fast process)\n- [MIMIC-CXR JPG] (https://physionet.org/content/mimic-cxr-jpg/2.0.0/) (Requires submission for access but its a pretty fast process)\n- [ChexPert](https://stanfordmlgroup.github.io/competitions/chexpert/) (Requires submission. There are external providers too including [tf](https://www.tensorflow.org/datasets/catalog/chexpert) and [small kaggle](https://www.kaggle.com/awsaf49/chexpert-dataset))\n- https://github.com/ieee8023/covid-chestxray-dataset\n\nBonus: [COVID-Net] (https://github.com/lindawangg/COVID-Net)", "1322922": "FYI, COVID-Net data can also be found in this [kaggle dataset](https://www.kaggle.com/andyczhao/covidx-cxr2).\n\nI'm just going to shamelessly advertised for COVID-Net as it came out from my lab. :)", "3369036": "i need to use the dataset MIMIC-CXR JPG. pls guide me how to use that. what are the procedures to use", "1386522": "Hi,\nAre public Kaggle data like\n* https://www.kaggle.com/tawsifurrahman/covid19-radiography-database\n* https://www.kaggle.com/andyczhao/covidx-cxr2\n* https://www.kaggle.com/preetviradiya/covid19-radiography-dataset\n* https://www.kaggle.com/prashant268/chest-xray-covid19-pneumonia\n* https://www.kaggle.com/paultimothymooney/chest-xray-pneumonia\n* https://www.kaggle.com/praveengovi/coronahack-chest-xraydataset\n* https://www.kaggle.com/pcbreviglieri/pneumonia-xray-images\n* https://www.kaggle.com/tolgadincer/labeled-chest-xray-images\n* https://www.kaggle.com/unaissait/curated-chest-xray-image-dataset-for-covid19\n* https://www.kaggle.com/edoardovantaggiato/covid19-xray-two-proposed-databases\n* https://www.kaggle.com/endiqq/largest-covid19-dataset\nalso available?\nThanks!", "1331795": "I'm sorry if I'm wrong, but isn't the mimic data only available to researchers?\nEven if not, we can't download the data without taking a test. Do we have permission to use the mimic data?", "1328152": "", "1328147": "" } }