I want to open \".dcm\" files with \"tensorflow-io\" functions but I found it's not installed on Kaggle kernels by default. To install in manually I need to have internet access which is forbidden for this competition.
\nthe first question is why internet access is disabled? is there any specific reason behind it?
is there any solution to have \"tenserflow-io\" without internet access?
or any suggestion to work with \".dcm\" files with another library or module?
I want to open \".dcm\" files with \"tensorflow-io\" functions but I found it's not installed on Kaggle kernels by default. To install in manually I need to have internet access which is forbidden for this competition.
\nthe first question is why internet access is disabled? is there any specific reason behind it?
is there any solution to have \"tenserflow-io\" without internet access?
or any suggestion to work with \".dcm\" files with another library or module?
Internet access is turned off for contests with a hidden dataset. Otherwise, you could leak the hidden dataset to an internet site.
\nAs explained elsewhere, generally you need to put the package you need into a dataset and load it from the dataset.
\n-Rich
", "rawMarkdown": "Internet access is turned off for contests with a hidden dataset. Otherwise, you could leak the hidden dataset to an internet site.\n\nAs explained elsewhere, generally you need to put the package you need into a dataset and load it from the dataset.\n\n-Rich", "votes": 3 }, { "id": 1437217, "postDate": "2021-08-03T15:28:40.060Z", "content": "It looks like it would be possible to download a copy of the install package and put it in a kaggle dataset. Here is an example of a pytorch lighting package. You would need to include all the dependencies, so it may take some experimentation. The main downside is that this dataset would not be updated automatically, so it can get out of date soon.
", "rawMarkdown": "It looks like it would be possible to download a copy of the install package and put it in a kaggle dataset. Here is an example of [a pytorch lighting package](https://www.kaggle.com/higepon/pytorchlightning-071). You would need to include all the dependencies, so it may take some experimentation. The main downside is that this dataset would not be updated automatically, so it can get out of date soon.", "votes": 2 }, { "id": 1430807, "postDate": "2021-08-03T11:55:50.123Z", "content": "!pip install tensorflow-ioHope that helps you with your questions. Good luck with the competition.
", "rawMarkdown": "1. The rules of the competition state that internet access is disabled. See the following [link](https://www.kaggle.com/c/rsna-miccai-brain-tumor-radiogenomic-classification/overview/code-requirements). Not sure what the particular reason is for this competition. Might limit the usefulness of tensorflow-io depending on what you are aiming to do with it.\n2. You can install tensorflow-io on Kaggle (I just tried) using the following code\n`!pip install tensorflow-io`\nIt may pop up some errors about dependencies but it looks like the install works fine. Up to you if you want to fix the dependency issues. \n3. Most of the notebooks that I have seen so far have been using pydicom for handling .dcm files. I think that it is already part of the base install.\n\nHope that helps you with your questions. Good luck with the competition.", "replies": [ { "id": 1433512, "postDate": "2021-08-03T14:03:59.637Z", "content": "Thanks for your comment
\nI've tried !pip install tensorflow-io but it works when internet is ON, that is not allowed in this competition.
but thank you for offering pydicom. I will try it.
", "rawMarkdown": "Thanks for your comment\n\nI've tried `!pip install tensorflow-io` but it works when internet is ON, that is not allowed in this competition.\n\nbut thank you for offering pydicom. I will try it.", "votes": 1 } ] }, { "id": 1430194, "postDate": "2021-08-03T11:23:51.233Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 1440295, "author_name": "quadcore/Richard Epstein", "author_url": "", "post_date": "2021-08-03T18:07:09.123000", "content": "Internet access is turned off for contests with a hidden dataset. Otherwise, you could leak the hidden dataset to an internet site.
\nAs explained elsewhere, generally you need to put the package you need into a dataset and load it from the dataset.
\n-Rich
", "votes": 3, "replies": [] }, { "id": 1437217, "author_name": "Bruce Harold", "author_url": "", "post_date": "2021-08-03T15:28:40.060000", "content": "It looks like it would be possible to download a copy of the install package and put it in a kaggle dataset. Here is an example of a pytorch lighting package. You would need to include all the dependencies, so it may take some experimentation. The main downside is that this dataset would not be updated automatically, so it can get out of date soon.
", "votes": 2, "replies": [] }, { "id": 1430807, "author_name": "Allan Bond", "author_url": "", "post_date": "2021-08-03T11:55:50.123000", "content": "!pip install tensorflow-ioHope that helps you with your questions. Good luck with the competition.
", "votes": 0, "replies": [ { "id": 1433512, "author_name": "Alireza Sadeghi", "author_url": "", "post_date": "2021-08-03T14:03:59.637000", "content": "Thanks for your comment
\nI've tried !pip install tensorflow-io but it works when internet is ON, that is not allowed in this competition.
but thank you for offering pydicom. I will try it.
", "votes": 1, "replies": [] } ] }, { "id": 1430194, "author_name": "", "author_url": "", "post_date": "2021-08-03T11:23:51.233000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "1429069": "I want to open \".dcm\" files with \"tensorflow-io\" functions but I found it's not installed on Kaggle kernels by default. To install in manually I need to have internet access which is forbidden for this competition.\n\n1. the first question is why internet access is disabled? is there any specific reason behind it?\n\n2. is there any solution to have \"tenserflow-io\" without internet access?\n\n3. or any suggestion to work with \".dcm\" files with another library or module?\n", "1440295": "Internet access is turned off for contests with a hidden dataset. Otherwise, you could leak the hidden dataset to an internet site.\n\nAs explained elsewhere, generally you need to put the package you need into a dataset and load it from the dataset.\n\n-Rich", "1437217": "It looks like it would be possible to download a copy of the install package and put it in a kaggle dataset. Here is an example of [a pytorch lighting package](https://www.kaggle.com/higepon/pytorchlightning-071). You would need to include all the dependencies, so it may take some experimentation. The main downside is that this dataset would not be updated automatically, so it can get out of date soon.", "1430807": "1. The rules of the competition state that internet access is disabled. See the following [link](https://www.kaggle.com/c/rsna-miccai-brain-tumor-radiogenomic-classification/overview/code-requirements). Not sure what the particular reason is for this competition. Might limit the usefulness of tensorflow-io depending on what you are aiming to do with it.\n2. You can install tensorflow-io on Kaggle (I just tried) using the following code\n`!pip install tensorflow-io`\nIt may pop up some errors about dependencies but it looks like the install works fine. Up to you if you want to fix the dependency issues. \n3. Most of the notebooks that I have seen so far have been using pydicom for handling .dcm files. I think that it is already part of the base install.\n\nHope that helps you with your questions. Good luck with the competition.", "1430194": "" } }