{ "id": 234575, "title": "WOW amazing! So many skilled Novice!", "url": "/competitions/hubmap-kidney-segmentation/discussion/234575", "author_name": "", "post_date": "2021-04-25T02:16:53.516260600Z", "votes": 26, "comment_count": 35, "views": 0, "content": "

I notice that this competition has so many new account joined in these few days.
\nIt is so nice to see so many new kaggler join this big community.
\nThe amazing part is, some people even got 0.934+ just in 3 days!

\n

SO NICE!
\nSO GREAT!
\nMAKE KAGGLE GREAT AGAIN :)

", "messages": [ { "id": "1283515", "postDate": "04/25/2021 02:16:53", "content": "

I notice that this competition has so many new account joined in these few days.
\nIt is so nice to see so many new kaggler join this big community.
\nThe amazing part is, some people even got 0.934+ just in 3 days!

\n

SO NICE!
\nSO GREAT!
\nMAKE KAGGLE GREAT AGAIN :)

", "rawMarkdown": "I notice that this competition has so many new account joined in these few days.\nIt is so nice to see so many new kaggler join this big community.\nThe amazing part is, some people even got 0.934+ just in 3 days!\n\nSO NICE! \nSO GREAT! \nMAKE KAGGLE GREAT AGAIN :)", "votes": null }, { "id": "1283641", "postDate": "04/25/2021 06:04:17", "content": "

I agree. Yeah, in these few days, so many people get amazing results. :)

", "rawMarkdown": "I agree. Yeah, in these few days, so many people get amazing results. :)", "votes": null }, { "id": "1283657", "postDate": "04/25/2021 06:30:16", "content": "

meanwhile in one day 10+ people crossed our score in LB , Is something intresting shared in the discussion that i am missing ?

", "rawMarkdown": "meanwhile in one day 10+ people crossed our score in LB , Is something intresting shared in the discussion that i am missing ?", "votes": null }, { "id": "1283675", "postDate": "04/25/2021 07:08:51", "content": "

maybe they are all geniuses and attempt time=3 equals to us=300.

", "rawMarkdown": "maybe they are all geniuses and attempt time=3 equals to us=300.", "votes": null }, { "id": "1284025", "postDate": "04/25/2021 13:55:38", "content": "

Thank you!! I certainly am! Hardly a novice in programming though. My current experience is in distributed ledgers

", "rawMarkdown": "Thank you!! I certainly am! Hardly a novice in programming though. My current experience is in distributed ledgers", "votes": null }, { "id": "1284027", "postDate": "04/25/2021 13:57:17", "content": "

I noticed that too. People passing by me at incredible speed. My rank dropped by about 20 in the past 3 days. I think it's the hand labeling, or cooperation?

", "rawMarkdown": "I noticed that too. People passing by me at incredible speed. My rank dropped by about 20 in the past 3 days. I think it's the hand labeling, or cooperation?", "votes": null }, { "id": "1284537", "postDate": "04/26/2021 04:07:29", "content": "

Well there is a public inference notebook in the 0.93x range. The models are private but if people use the output submission file only maybe hand label or post process to tweak or blend, it could get a better public LB score.

\n

I worry if new kagglers do not realise that the true submission to get private test and LB score cannot be done this way.
\nMaybe a topic is needed for how to do final submissions.

\n

But this competition is more accessible than most computer vision ones in terms of size of datasets, models that can be used, so hopefully that has allowed more to join. Best of luck to all here.

", "rawMarkdown": "Well there is a public inference notebook in the 0.93x range. The models are private but if people use the output submission file only maybe hand label or post process to tweak or blend, it could get a better public LB score. \n\nI worry if new kagglers do not realise that the true submission to get private test and LB score cannot be done this way. \nMaybe a topic is needed for how to do final submissions. \n\nBut this competition is more accessible than most computer vision ones in terms of size of datasets, models that can be used, so hopefully that has allowed more to join. Best of luck to all here.", "votes": null }, { "id": "1284976", "postDate": "04/26/2021 13:15:47", "content": "

Some might be people circumventing the 5 submissions per day restriction. πŸ€”

", "rawMarkdown": "Some might be people circumventing the 5 submissions per day restriction. πŸ€”", "votes": null }, { "id": "1284998", "postDate": "04/26/2021 13:36:09", "content": "

or better yet-the 30 hour TPU limit

", "rawMarkdown": "or better yet-the 30 hour TPU limit", "votes": null }, { "id": "1285035", "postDate": "04/26/2021 14:10:21", "content": "

haha, i`m exactly the same πŸ˜…

", "rawMarkdown": "haha, i`m exactly the same πŸ˜…", "votes": null }, { "id": "1285347", "postDate": "04/26/2021 20:12:08", "content": "

@rosuluc , we are originally from the same country too!

", "rawMarkdown": "rosuluc , we are originally from the same country too!", "votes": null }, { "id": "1286618", "postDate": "04/28/2021 07:57:54", "content": "

\"lol.png\"

\n

2 subs πŸ‘€
\nLooks like kaggle is gonna do some cleaning

", "rawMarkdown": "[![lol.png](https://i.postimg.cc/5tgfmTmp/lol.png)](https://postimg.cc/677sWHgG)\n\n2 subs πŸ‘€\nLooks like kaggle is gonna do some cleaning", "votes": null }, { "id": "1286803", "postDate": "04/28/2021 12:21:27", "content": "

Here is another one, joined yesterday and now at 20th position.

\n

\"\"

", "rawMarkdown": "Here is another one, joined yesterday and now at 20th position.\n\n![](https://i.postimg.cc/nh7vX8X7/Screenshot-2021-04-28-at-5-46-08-PM.png)", "votes": null }, { "id": "1287119", "postDate": "04/28/2021 18:02:30", "content": "

Hope will disappear soon.

", "rawMarkdown": "Hope will disappear soon.", "votes": null }, { "id": "1287121", "postDate": "04/28/2021 18:03:45", "content": "

It is not possible even for an expert to join kaggle and reach rank 20 in 1 day.

", "rawMarkdown": "It is not possible even for an expert to join kaggle and reach rank 20 in 1 day.", "votes": null }, { "id": "1287465", "postDate": "04/29/2021 05:42:43", "content": "
    \n
  1. Kaggle Novice doesn't mean I have never joined other competitions on other platforms.
    2. \n
  2. I just copy Matthias Griebel's deepflash2 notebook ,use pseudo labels by Zhao, and tune the params. You can try his notebook, really fast.
    \nrelax brosπŸ˜‚
    3. \n
", "rawMarkdown": "1. Kaggle Novice doesn't mean I have never joined other competitions on other platforms.\n2. I just copy Matthias Griebel's deepflash2 notebook ,use pseudo labels by Zhao, and tune the params. You can try his notebook, really fast.\nrelax brosπŸ˜‚", "votes": null }, { "id": "1287470", "postDate": "04/29/2021 05:48:43", "content": "

Yup!πŸ˜… Kaggle Novice does not mean new to data science. Can you tell me what is your CV which you have reached using above method ?

", "rawMarkdown": "Yup!πŸ˜… Kaggle Novice does not mean new to data science. Can you tell me what is your CV which you have reached using above method ?", "votes": null }, { "id": "1287502", "postDate": "04/29/2021 06:20:56", "content": "

About 95.3. I think his sample way may cause leak. Maybe I am wrong since I haven't understood the deepflash2 function.

", "rawMarkdown": "About 95.3. I think his sample way may cause leak. Maybe I am wrong since I haven't understood the deepflash2 function.", "votes": null }, { "id": "1287507", "postDate": "04/29/2021 06:25:10", "content": "

Sooner kaggle will organize a competition to detect the kagglers gaining ranks/medals by illegitimate means..

", "rawMarkdown": "Sooner kaggle will organize a competition to detect the kagglers gaining ranks/medals by illegitimate means..", "votes": null }, { "id": "1288531", "postDate": "04/30/2021 05:19:28", "content": "

This is very painful, I tried to take part in the competition but could not because I was even unable to organize and use the data properly and seeing this I am sad. I can very well understand how the masters and grandmasters are feeling when they work for months and suddenly something like this keeps happening. I hope kaggle takes some steps.

\n

Also I would like to take this opportunity to ask can you guys share some resources that will help me get better in understanding image segmentation.

", "rawMarkdown": "This is very painful, I tried to take part in the competition but could not because I was even unable to organize and use the data properly and seeing this I am sad. I can very well understand how the masters and grandmasters are feeling when they work for months and suddenly something like this keeps happening. I hope kaggle takes some steps. \n\nAlso I would like to take this opportunity to ask can you guys share some resources that will help me get better in understanding image segmentation.", "votes": null }, { "id": "1288592", "postDate": "04/30/2021 06:34:07", "content": "

I think they might go to Harvard.
\n

", "rawMarkdown": "I think they might go to Harvard. \n", "votes": null }, { "id": "1288629", "postDate": "04/30/2021 07:17:25", "content": "

There are 2 accounts in the top 30 that joined on the same day, are participating in the same 2 competitions and have identical rankings. One hell of a coincidence. I don't want to start a witch hunt. But there is something fishy about this competition.

", "rawMarkdown": "There are 2 accounts in the top 30 that joined on the same day, are participating in the same 2 competitions and have identical rankings. One hell of a coincidence. I don't want to start a witch hunt. But there is something fishy about this competition.", "votes": null }, { "id": "1288837", "postDate": "04/30/2021 11:45:07", "content": "

Also submitted at the same time, what a coincidence :)

", "rawMarkdown": "Also submitted at the same time, what a coincidence :)", "votes": null }, { "id": "1288853", "postDate": "04/30/2021 12:06:41", "content": "

when algorithms are shared, then it is more relying on hardware differences.

", "rawMarkdown": "when algorithms are shared, then it is more relying on hardware differences.", "votes": null }, { "id": "1288857", "postDate": "04/30/2021 12:12:15", "content": "

πŸ˜‚πŸ˜‚amazing

", "rawMarkdown": "πŸ˜‚πŸ˜‚amazing", "votes": null }, { "id": "1289021", "postDate": "04/30/2021 15:09:02", "content": "

So in less than a day after joining Kaggle you were able to :

\n\n

I'll let the Kaggle team draw the conclusions but I'm not buying it :)

", "rawMarkdown": "So in less than a day after **joining** Kaggle you were able to :\n- Understand that Zhao's labels were the key to climb LB (you have to follow the forum to get such info)\n- Tune the deepflash notebook enough to get to top 30 - provided it takes a few hours to train a model and a few more to submit that's impressive.\n\nI'll let the Kaggle team draw the conclusions but I'm not buying it :)", "votes": null }, { "id": "1289074", "postDate": "04/30/2021 16:04:31", "content": "

that's why I don't understand why people have to make their notebooks public before the competition has closed!

", "rawMarkdown": "that's why I don't understand why people have to make their notebooks public before the competition has closed!", "votes": null }, { "id": "1289406", "postDate": "05/01/2021 00:06:06", "content": "

It took me a little while to find out that the labels were the key and I am someone who's been avidly reading the forums everyday.

", "rawMarkdown": "It took me a little while to find out that the labels were the key and I am someone who's been avidly reading the forums everyday.", "votes": null }, { "id": "1290530", "postDate": "05/02/2021 05:07:34", "content": "

Hello,where is Zhao's label?

", "rawMarkdown": "Hello,where is Zhao's label?", "votes": null }, { "id": "1290571", "postDate": "05/02/2021 06:24:56", "content": "

https://www.kaggle.com/carnozhao/d48-hand-labelled

", "rawMarkdown": "[https://www.kaggle.com/carnozhao/d48-hand-labelled](url)", "votes": null }, { "id": "1290994", "postDate": "05/02/2021 16:01:31", "content": "

Hey @forcewithme Question about the D48…How do you use it then?

\n
    \n
  1. Do you use it to retrain a model?
    2. \n
  2. Or only use it to modify your final submission?
    3. \n
", "rawMarkdown": "Hey @forcewithme Question about the D48...How do you use it then?\n\n1. Do you use it to retrain a model?\n2. Or only use it to modify your final submission?", "votes": null }, { "id": "1291074", "postDate": "05/02/2021 17:19:09", "content": "

I use it to retrain the model.
\nI copy deepflash2's notebook to make D4 patch in zarr, and add the D4 patch into deepflash2's dataset.

", "rawMarkdown": "I use it to retrain the model.\nI copy deepflash2's notebook to make D4 patch in zarr, and add the D4 patch into deepflash2's dataset.", "votes": null }, { "id": "1291107", "postDate": "05/02/2021 17:40:57", "content": "

Thanks @forcewithme for clarifying :-)

", "rawMarkdown": "Thanks @forcewithme for clarifying :-)", "votes": null }, { "id": "1299817", "postDate": "05/10/2021 04:10:37", "content": "

Now both of these accounts have merged and are in the top 20. Wow @forcewithme @oldpotato care to explain?

", "rawMarkdown": "Now both of these accounts have merged and are in the top 20. Wow @forcewithme @oldpotato care to explain?", "votes": null }, { "id": "1299941", "postDate": "05/10/2021 06:49:35", "content": "

I don't even want to point that out, it is way too obvious. They might assume people are blind.

", "rawMarkdown": "I don't even want to point that out, it is way too obvious. They might assume people are blind.", "votes": null }, { "id": "1300023", "postDate": "05/10/2021 08:06:34", "content": "

Worst part is that this is not one competition but both the competitions, these accounts have the same score and submit within an hour of each other. I mean you can't get more obvious than that.

", "rawMarkdown": "Worst part is that this is not one competition but both the competitions, these accounts have the same score and submit within an hour of each other. I mean you can't get more obvious than that.", "votes": null } ], "comments": [ { "id": 1283641, "author_name": "shinewine", "author_url": "", "post_date": "04/25/2021 06:04:17", "content": "

I agree. Yeah, in these few days, so many people get amazing results. :)

", "votes": null, "replies": [] }, { "id": 1283657, "author_name": "trooperog", "author_url": "", "post_date": "04/25/2021 06:30:16", "content": "

meanwhile in one day 10+ people crossed our score in LB , Is something intresting shared in the discussion that i am missing ?

", "votes": null, "replies": [ { "id": 1284027, "author_name": "andrasferenczi", "author_url": "", "post_date": "04/25/2021 13:57:17", "content": "

I noticed that too. People passing by me at incredible speed. My rank dropped by about 20 in the past 3 days. I think it's the hand labeling, or cooperation?

", "votes": null, "replies": [] } ] }, { "id": 1283675, "author_name": "southsakura", "author_url": "", "post_date": "04/25/2021 07:08:51", "content": "

maybe they are all geniuses and attempt time=3 equals to us=300.

", "votes": null, "replies": [] }, { "id": 1284025, "author_name": "andrasferenczi", "author_url": "", "post_date": "04/25/2021 13:55:38", "content": "

Thank you!! I certainly am! Hardly a novice in programming though. My current experience is in distributed ledgers

", "votes": null, "replies": [ { "id": 1285035, "author_name": "rosuluc", "author_url": "", "post_date": "04/26/2021 14:10:21", "content": "

haha, i`m exactly the same πŸ˜…

", "votes": null, "replies": [] }, { "id": 1285347, "author_name": "andrasferenczi", "author_url": "", "post_date": "04/26/2021 20:12:08", "content": "

@rosuluc , we are originally from the same country too!

", "votes": null, "replies": [] } ] }, { "id": 1284537, "author_name": "something4kag", "author_url": "", "post_date": "04/26/2021 04:07:29", "content": "

Well there is a public inference notebook in the 0.93x range. The models are private but if people use the output submission file only maybe hand label or post process to tweak or blend, it could get a better public LB score.

\n

I worry if new kagglers do not realise that the true submission to get private test and LB score cannot be done this way.
\nMaybe a topic is needed for how to do final submissions.

\n

But this competition is more accessible than most computer vision ones in terms of size of datasets, models that can be used, so hopefully that has allowed more to join. Best of luck to all here.

", "votes": null, "replies": [] }, { "id": 1284976, "author_name": "erikdali", "author_url": "", "post_date": "04/26/2021 13:15:47", "content": "

Some might be people circumventing the 5 submissions per day restriction. πŸ€”

", "votes": null, "replies": [ { "id": 1284998, "author_name": "andrasferenczi", "author_url": "", "post_date": "04/26/2021 13:36:09", "content": "

or better yet-the 30 hour TPU limit

", "votes": null, "replies": [] } ] }, { "id": 1286618, "author_name": "theoviel", "author_url": "", "post_date": "04/28/2021 07:57:54", "content": "

\"lol.png\"

\n

2 subs πŸ‘€
\nLooks like kaggle is gonna do some cleaning

", "votes": null, "replies": [ { "id": 1286803, "author_name": "rohitsingh9990", "author_url": "", "post_date": "04/28/2021 12:21:27", "content": "

Here is another one, joined yesterday and now at 20th position.

\n

\"\"

", "votes": null, "replies": [] }, { "id": 1287119, "author_name": "adg1822", "author_url": "", "post_date": "04/28/2021 18:02:30", "content": "

Hope will disappear soon.

", "votes": null, "replies": [] }, { "id": 1287121, "author_name": "adg1822", "author_url": "", "post_date": "04/28/2021 18:03:45", "content": "

It is not possible even for an expert to join kaggle and reach rank 20 in 1 day.

", "votes": null, "replies": [] }, { "id": 1287465, "author_name": "forcewithme", "author_url": "", "post_date": "04/29/2021 05:42:43", "content": "
    \n
  1. Kaggle Novice doesn't mean I have never joined other competitions on other platforms.
    2. \n
  2. I just copy Matthias Griebel's deepflash2 notebook ,use pseudo labels by Zhao, and tune the params. You can try his notebook, really fast.
    \nrelax brosπŸ˜‚
    3. \n
", "votes": null, "replies": [] }, { "id": 1287470, "author_name": "adg1822", "author_url": "", "post_date": "04/29/2021 05:48:43", "content": "

Yup!πŸ˜… Kaggle Novice does not mean new to data science. Can you tell me what is your CV which you have reached using above method ?

", "votes": null, "replies": [] }, { "id": 1287502, "author_name": "forcewithme", "author_url": "", "post_date": "04/29/2021 06:20:56", "content": "

About 95.3. I think his sample way may cause leak. Maybe I am wrong since I haven't understood the deepflash2 function.

", "votes": null, "replies": [] }, { "id": 1289021, "author_name": "theoviel", "author_url": "", "post_date": "04/30/2021 15:09:02", "content": "

So in less than a day after joining Kaggle you were able to :

\n\n

I'll let the Kaggle team draw the conclusions but I'm not buying it :)

", "votes": null, "replies": [] }, { "id": 1289406, "author_name": "dragonpg2000", "author_url": "", "post_date": "05/01/2021 00:06:06", "content": "

It took me a little while to find out that the labels were the key and I am someone who's been avidly reading the forums everyday.

", "votes": null, "replies": [] }, { "id": 1290530, "author_name": "zekunn", "author_url": "", "post_date": "05/02/2021 05:07:34", "content": "

Hello,where is Zhao's label?

", "votes": null, "replies": [] }, { "id": 1290571, "author_name": "forcewithme", "author_url": "", "post_date": "05/02/2021 06:24:56", "content": "

https://www.kaggle.com/carnozhao/d48-hand-labelled

", "votes": null, "replies": [] }, { "id": 1290994, "author_name": "rsmits", "author_url": "", "post_date": "05/02/2021 16:01:31", "content": "

Hey @forcewithme Question about the D48…How do you use it then?

\n
    \n
  1. Do you use it to retrain a model?
    2. \n
  2. Or only use it to modify your final submission?
    3. \n
", "votes": null, "replies": [] }, { "id": 1291074, "author_name": "forcewithme", "author_url": "", "post_date": "05/02/2021 17:19:09", "content": "

I use it to retrain the model.
\nI copy deepflash2's notebook to make D4 patch in zarr, and add the D4 patch into deepflash2's dataset.

", "votes": null, "replies": [] }, { "id": 1291107, "author_name": "rsmits", "author_url": "", "post_date": "05/02/2021 17:40:57", "content": "

Thanks @forcewithme for clarifying :-)

", "votes": null, "replies": [] }, { "id": 1299817, "author_name": "dragonpg2000", "author_url": "", "post_date": "05/10/2021 04:10:37", "content": "

Now both of these accounts have merged and are in the top 20. Wow @forcewithme @oldpotato care to explain?

", "votes": null, "replies": [] }, { "id": 1299941, "author_name": "xiejialun", "author_url": "", "post_date": "05/10/2021 06:49:35", "content": "

I don't even want to point that out, it is way too obvious. They might assume people are blind.

", "votes": null, "replies": [] }, { "id": 1300023, "author_name": "dragonpg2000", "author_url": "", "post_date": "05/10/2021 08:06:34", "content": "

Worst part is that this is not one competition but both the competitions, these accounts have the same score and submit within an hour of each other. I mean you can't get more obvious than that.

", "votes": null, "replies": [] } ] }, { "id": 1287507, "author_name": "jaideepvalani", "author_url": "", "post_date": "04/29/2021 06:25:10", "content": "

Sooner kaggle will organize a competition to detect the kagglers gaining ranks/medals by illegitimate means..

", "votes": null, "replies": [] }, { "id": 1288531, "author_name": "sayantansadhu", "author_url": "", "post_date": "04/30/2021 05:19:28", "content": "

This is very painful, I tried to take part in the competition but could not because I was even unable to organize and use the data properly and seeing this I am sad. I can very well understand how the masters and grandmasters are feeling when they work for months and suddenly something like this keeps happening. I hope kaggle takes some steps.

\n

Also I would like to take this opportunity to ask can you guys share some resources that will help me get better in understanding image segmentation.

", "votes": null, "replies": [] }, { "id": 1288592, "author_name": "underwearfitting", "author_url": "", "post_date": "04/30/2021 06:34:07", "content": "

I think they might go to Harvard.
\n

", "votes": null, "replies": [] }, { "id": 1288629, "author_name": "dragonpg2000", "author_url": "", "post_date": "04/30/2021 07:17:25", "content": "

There are 2 accounts in the top 30 that joined on the same day, are participating in the same 2 competitions and have identical rankings. One hell of a coincidence. I don't want to start a witch hunt. But there is something fishy about this competition.

", "votes": null, "replies": [ { "id": 1288837, "author_name": "xiejialun", "author_url": "", "post_date": "04/30/2021 11:45:07", "content": "

Also submitted at the same time, what a coincidence :)

", "votes": null, "replies": [] }, { "id": 1288857, "author_name": "adg1822", "author_url": "", "post_date": "04/30/2021 12:12:15", "content": "

πŸ˜‚πŸ˜‚amazing

", "votes": null, "replies": [] } ] }, { "id": 1288853, "author_name": "dragonzhang", "author_url": "", "post_date": "04/30/2021 12:06:41", "content": "

when algorithms are shared, then it is more relying on hardware differences.

", "votes": null, "replies": [ { "id": 1289074, "author_name": "erikdali", "author_url": "", "post_date": "04/30/2021 16:04:31", "content": "

that's why I don't understand why people have to make their notebooks public before the competition has closed!

", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "1283515": "I notice that this competition has so many new account joined in these few days.\nIt is so nice to see so many new kaggler join this big community.\nThe amazing part is, some people even got 0.934+ just in 3 days!\n\nSO NICE! \nSO GREAT! \nMAKE KAGGLE GREAT AGAIN :)", "1283641": "I agree. Yeah, in these few days, so many people get amazing results. :)", "1283657": "meanwhile in one day 10+ people crossed our score in LB , Is something intresting shared in the discussion that i am missing ?", "1283675": "maybe they are all geniuses and attempt time=3 equals to us=300.", "1284025": "Thank you!! I certainly am! Hardly a novice in programming though. My current experience is in distributed ledgers", "1284027": "I noticed that too. People passing by me at incredible speed. My rank dropped by about 20 in the past 3 days. I think it's the hand labeling, or cooperation?", "1284537": "Well there is a public inference notebook in the 0.93x range. The models are private but if people use the output submission file only maybe hand label or post process to tweak or blend, it could get a better public LB score. \n\nI worry if new kagglers do not realise that the true submission to get private test and LB score cannot be done this way. \nMaybe a topic is needed for how to do final submissions. \n\nBut this competition is more accessible than most computer vision ones in terms of size of datasets, models that can be used, so hopefully that has allowed more to join. Best of luck to all here.", "1284976": "Some might be people circumventing the 5 submissions per day restriction. πŸ€”", "1284998": "or better yet-the 30 hour TPU limit", "1285035": "haha, i`m exactly the same πŸ˜…", "1285347": "rosuluc , we are originally from the same country too!", "1286618": "[![lol.png](https://i.postimg.cc/5tgfmTmp/lol.png)](https://postimg.cc/677sWHgG)\n\n2 subs πŸ‘€\nLooks like kaggle is gonna do some cleaning", "1286803": "Here is another one, joined yesterday and now at 20th position.\n\n![](https://i.postimg.cc/nh7vX8X7/Screenshot-2021-04-28-at-5-46-08-PM.png)", "1287119": "Hope will disappear soon.", "1287121": "It is not possible even for an expert to join kaggle and reach rank 20 in 1 day.", "1287465": "1. Kaggle Novice doesn't mean I have never joined other competitions on other platforms.\n2. I just copy Matthias Griebel's deepflash2 notebook ,use pseudo labels by Zhao, and tune the params. You can try his notebook, really fast.\nrelax brosπŸ˜‚", "1287470": "Yup!πŸ˜… Kaggle Novice does not mean new to data science. Can you tell me what is your CV which you have reached using above method ?", "1287502": "About 95.3. I think his sample way may cause leak. Maybe I am wrong since I haven't understood the deepflash2 function.", "1287507": "Sooner kaggle will organize a competition to detect the kagglers gaining ranks/medals by illegitimate means..", "1288531": "This is very painful, I tried to take part in the competition but could not because I was even unable to organize and use the data properly and seeing this I am sad. I can very well understand how the masters and grandmasters are feeling when they work for months and suddenly something like this keeps happening. I hope kaggle takes some steps. \n\nAlso I would like to take this opportunity to ask can you guys share some resources that will help me get better in understanding image segmentation.", "1288592": "I think they might go to Harvard. \n", "1288629": "There are 2 accounts in the top 30 that joined on the same day, are participating in the same 2 competitions and have identical rankings. One hell of a coincidence. I don't want to start a witch hunt. But there is something fishy about this competition.", "1288837": "Also submitted at the same time, what a coincidence :)", "1288853": "when algorithms are shared, then it is more relying on hardware differences.", "1288857": "πŸ˜‚πŸ˜‚amazing", "1289021": "So in less than a day after **joining** Kaggle you were able to :\n- Understand that Zhao's labels were the key to climb LB (you have to follow the forum to get such info)\n- Tune the deepflash notebook enough to get to top 30 - provided it takes a few hours to train a model and a few more to submit that's impressive.\n\nI'll let the Kaggle team draw the conclusions but I'm not buying it :)", "1289074": "that's why I don't understand why people have to make their notebooks public before the competition has closed!", "1289406": "It took me a little while to find out that the labels were the key and I am someone who's been avidly reading the forums everyday.", "1290530": "Hello,where is Zhao's label?", "1290571": "[https://www.kaggle.com/carnozhao/d48-hand-labelled](url)", "1290994": "Hey @forcewithme Question about the D48...How do you use it then?\n\n1. Do you use it to retrain a model?\n2. Or only use it to modify your final submission?", "1291074": "I use it to retrain the model.\nI copy deepflash2's notebook to make D4 patch in zarr, and add the D4 patch into deepflash2's dataset.", "1291107": "Thanks @forcewithme for clarifying :-)", "1299817": "Now both of these accounts have merged and are in the top 20. Wow @forcewithme @oldpotato care to explain?", "1299941": "I don't even want to point that out, it is way too obvious. They might assume people are blind.", "1300023": "Worst part is that this is not one competition but both the competitions, these accounts have the same score and submit within an hour of each other. I mean you can't get more obvious than that." }, "source": "meta" }