Hello friends. I am a beginner. And this is the first competition I'm participating in. I have 3 questions. my questions may be laughable to you!\n1. Where and how is mask data used? Shouldn't we train the network with train_images data? So what is mask data for ?\n2. Do we classify based on isup_grade? So what is gleason_score for?\n3- How can I delete the white parts of the image? These sections cover a large part of the image and make the training time consuming.\nThank you for If you help me.
", "messages": [ { "id": 840552, "postDate": "2020-05-10T06:14:58.430Z", "content": "All great questions, for 3 you can check https://www.kaggle.com/iafoss/panda-16x128x128-tiles as suggested by @zhangeng .
\n\nFor 1, so far top teams are not using masks it seems, but maybe someone will find a way to make them useful.\nFor 2 - yes we classify based on isup_grade, I'm personally didn't try to use gleason_score yet, but in theory it could be useful as an auxiliary target, thanks for a reminder, I already forgot it's available.
thanks a lot\nThank you for your useful information and devoting your time to answering my question.\nMany ambiguities in my mind were resolved.
", "rawMarkdown": "thanks a lot\nThank you for your useful information and devoting your time to answering my question.\nMany ambiguities in my mind were resolved.", "votes": 1 } ] }, { "id": 840441, "postDate": "2020-05-10T03:19:43.253Z", "content": "2.https://www.kaggle.com/iafoss/panda-16x128x128-tiles
", "rawMarkdown": "2.[https://www.kaggle.com/iafoss/panda-16x128x128-tiles](https://www.kaggle.com/iafoss/panda-16x128x128-tiles)", "replies": [ { "id": 842361, "postDate": "2020-05-11T11:04:48.723Z", "content": "Thank you very much really\nthank you for devoting your time to answering my question.
", "rawMarkdown": "Thank you very much really\nthank you for devoting your time to answering my question." } ] }, { "id": 840425, "postDate": "2020-05-10T02:36:16.657Z", "content": "Hello friends. I am a beginner. And this is the first competition I'm participating in. I have 3 questions. my questions may be laughable to you!\n1. Where and how is mask data used? Shouldn't we train the network with train_images data? So what is mask data for ?\n2. Do we classify based on isup_grade? So what is gleason_score for?\n3- How can I delete the white parts of the image? These sections cover a large part of the image and make the training time consuming.\nThank you for If you help me.
", "rawMarkdown": "Hello friends. I am a beginner. And this is the first competition I'm participating in. I have 3 questions. my questions may be laughable to you!\n1. Where and how is mask data used? Shouldn't we train the network with train_images data? So what is mask data for ?\n2. Do we classify based on isup_grade? So what is gleason_score for?\n3- How can I delete the white parts of the image? These sections cover a large part of the image and make the training time consuming.\nThank you for If you help me." } ], "comments": [ { "id": 840552, "author_name": "Konstantin Lopukhin", "author_url": "", "post_date": "2020-05-10T06:14:58.430000", "content": "All great questions, for 3 you can check https://www.kaggle.com/iafoss/panda-16x128x128-tiles as suggested by @zhangeng .
\n\nFor 1, so far top teams are not using masks it seems, but maybe someone will find a way to make them useful.\nFor 2 - yes we classify based on isup_grade, I'm personally didn't try to use gleason_score yet, but in theory it could be useful as an auxiliary target, thanks for a reminder, I already forgot it's available.
thanks a lot\nThank you for your useful information and devoting your time to answering my question.\nMany ambiguities in my mind were resolved.
", "votes": 1, "replies": [] } ] }, { "id": 840441, "author_name": "zhangeng", "author_url": "", "post_date": "2020-05-10T03:19:43.253000", "content": "2.https://www.kaggle.com/iafoss/panda-16x128x128-tiles
", "votes": 0, "replies": [ { "id": 842361, "author_name": "mohamad", "author_url": "", "post_date": "2020-05-11T11:04:48.723000", "content": "Thank you very much really\nthank you for devoting your time to answering my question.
", "votes": 0, "replies": [] } ] } ], "raw_markdown_by_id": { "840552": "All great questions, for 3 you can check https://www.kaggle.com/iafoss/panda-16x128x128-tiles as suggested by @zhangeng .\n\nFor 1, so far top teams are not using masks it seems, but maybe someone will find a way to make them useful.\nFor 2 - yes we classify based on ``isup_grade``, I'm personally didn't try to use ``gleason_score`` yet, but in theory it could be useful as an auxiliary target, thanks for a reminder, I already forgot it's available.", "840441": "2.[https://www.kaggle.com/iafoss/panda-16x128x128-tiles](https://www.kaggle.com/iafoss/panda-16x128x128-tiles)", "840425": "Hello friends. I am a beginner. And this is the first competition I'm participating in. I have 3 questions. my questions may be laughable to you!\n1. Where and how is mask data used? Shouldn't we train the network with train_images data? So what is mask data for ?\n2. Do we classify based on isup_grade? So what is gleason_score for?\n3- How can I delete the white parts of the image? These sections cover a large part of the image and make the training time consuming.\nThank you for If you help me." } }