{ "id": 417668, "title": "Update on acceptable licenses for datasets/pre-trained models", "url": "/competitions/benetech-making-graphs-accessible/discussion/417668", "author_name": "Ron Ellis", "post_date": "2023-06-16T16:57:05.624000", "votes": -27, "comment_count": 57, "views": 0, "content": "

I want to apologize for any confusion participants may have experienced as a result of rule changes that occurred.

\n

Out of respect for the hard work all of the participants have invested in their entries and in the spirit of fairness, we will NOT be disqualifying or otherwise penalizing participants due to use of datasets or pre-trained models licensed under any of the following:

\n\n

Thank you again to all of our participants for your hard work on this difficult problem. Please know we have been working diligently behind the scenes to ensure your valuable time and effort working with this data will not go to waste.

", "messages": [ { "id": 2305601, "postDate": "2023-06-16T19:42:21.063Z", "content": "

A punch in the face for all that played fair…

", "rawMarkdown": "A punch in the face for all that played fair...", "votes": 14, "replies": [ { "id": 2305618, "postDate": "2023-06-16T19:56:45.987Z", "content": "

We have a full MIT compliant solution. We spent countless days to make it work after the latest rules change.

", "rawMarkdown": "We have a full MIT compliant solution. We spent countless days to make it work after the latest rules change.", "votes": 8, "replies": [ { "id": 2305621, "postDate": "2023-06-16T19:58:28.853Z", "content": "

We did the same, despite it not being the easiest or the most effective method.

", "rawMarkdown": "We did the same, despite it not being the easiest or the most effective method.", "votes": 3 } ] }, { "id": 2306057, "postDate": "2023-06-17T04:46:00.567Z", "content": "

Definitely Agree!

", "rawMarkdown": "Definitely Agree!" } ] }, { "id": 2306826, "postDate": "2023-06-17T16:05:37.400Z", "content": "

@maggiemd,

\n

I think it would be great if all future competition organizers explicitly said which licenses were acceptable at the beginning of the competition so we can avoid situations like this.

", "rawMarkdown": "@maggiemd,\n\nI think it would be great if all future competition organizers explicitly said which licenses were acceptable at the beginning of the competition so we can avoid situations like this. ", "votes": 10, "replies": [ { "id": 2306950, "postDate": "2023-06-17T17:44:56.577Z", "content": "

It was stated clearly in the rules. What was insane is that host recommended a dataset incompatible with the rules.

", "rawMarkdown": "It was stated clearly in the rules. What was insane is that host recommended a dataset incompatible with the rules.", "votes": 3, "replies": [ { "id": 2307016, "postDate": "2023-06-17T18:32:06.890Z", "content": "

The rules say

\n
\n
    \n
  1. SUBMISSION CODE REQUIREMENTS. Use of Open Source
    \n The following supersedes Section 8.C of the rules below.
    \n Use of Open Source. If open source code is used in the model to generate the Submission, then you must only use open source code licensed under an Open Source Initiative-approved license (see www.opensource.org) that in no event limits commercial use of such code or model containing or depending on such code. Reference was made to an outside data set, as an example; however, we have determined that example does not meet Sponsor’s open source requirements. For clarification, we have removed the reference to that outside data set. If you wish to use any training data set other than those provided on the Competition Website, it is your responsibility to verify that, given the possibility of financial remuneration, you have the necessary rights to use such other data set.
    2. \n
\n
\n
\n
\n

7C. External Data. You may use data other than the Competition Data (“External Data”) to develop and test your Submissions. However, you will ensure the External Data is publicly available and equally accessible to use by all participants of the Competition for purposes of the competition at no cost to the other participants. The ability to use External Data under this Section 7.C (External Data) does not limit your other obligations under these Competition Rules, including but not limited to Section 11 (Winners Obligations).

\n
\n
\n
\n

8C. Use of Open Source. Unless otherwise stated in the Specific Competition Rules above, if open source code is used in the model to generate the Submission, then you must only use open source code licensed under an Open Source Initiative-approved license (see www.opensource.org) that in no event limits commercial use of such code or model containing or depending on such code.

\n
\n

The AGPL-3.0 license is listed on opensource.org, so I think there could still be confusion without the hosts explicitly indicating allowed licenses

", "rawMarkdown": "The rules say\n> 3. SUBMISSION CODE REQUIREMENTS. Use of Open Source \nThe following supersedes Section 8.C of the rules below.\nUse of Open Source. If open source code is used in the model to generate the Submission, then you must only use open source code licensed under an Open Source Initiative-approved license (see www.opensource.org) that in no event limits commercial use of such code or model containing or depending on such code. Reference was made to an outside data set, as an example; however, we have determined that example does not meet Sponsor’s open source requirements. For clarification, we have removed the reference to that outside data set. If you wish to use any training data set other than those provided on the Competition Website, it is your responsibility to verify that, given the possibility of financial remuneration, you have the necessary rights to use such other data set. \n\n\n---\n\n> 7C. External Data. You may use data other than the Competition Data (“External Data”) to develop and test your Submissions. However, you will ensure the External Data is publicly available and equally accessible to use by all participants of the Competition for purposes of the competition at no cost to the other participants. The ability to use External Data under this Section 7.C (External Data) does not limit your other obligations under these Competition Rules, including but not limited to Section 11 (Winners Obligations). \n \n\n---\n\n> 8C. Use of Open Source. Unless otherwise stated in the Specific Competition Rules above, if open source code is used in the model to generate the Submission, then you must only use open source code licensed under an Open Source Initiative-approved license (see www.opensource.org) that in no event limits commercial use of such code or model containing or depending on such code.\n\nThe AGPL-3.0 license is listed on opensource.org, so I think there could still be confusion without the hosts explicitly indicating allowed licenses" }, { "id": 2307055, "postDate": "2023-06-17T19:18:36.867Z", "content": "
\n

that in no event limits commercial use of such code or model containing or depending on such code.

\n
\n

Are you saying GNU AGPL 3.0 meets this?

", "rawMarkdown": "> that in no event limits commercial use of such code or model containing or depending on such code. \n\nAre you saying GNU AGPL 3.0 meets this?" }, { "id": 2307070, "postDate": "2023-06-17T19:34:53.833Z", "content": "

I'm not a legal expert, but it seems ok to me

\n

https://www.tldrlegal.com/license/gnu-affero-general-public-license-v3-agpl-3-0

", "rawMarkdown": "I'm not a legal expert, but it seems ok to me\n\nhttps://www.tldrlegal.com/license/gnu-affero-general-public-license-v3-agpl-3-0" } ] } ] }, { "id": 2305591, "postDate": "2023-06-16T19:32:50.090Z", "content": "

After spending countless additional days to make our code MIT compliant, this decision (3 days before the end) comes as a complete shock to me. Is this a joke? Again?

", "rawMarkdown": "After spending countless additional days to make our code MIT compliant, this decision (3 days before the end) comes as a complete shock to me. Is this a joke? Again?", "votes": 10, "replies": [ { "id": 2305592, "postDate": "2023-06-16T19:34:22.143Z", "content": "

Same here, unbelievable…

", "rawMarkdown": "Same here, unbelievable…", "votes": 3 } ] }, { "id": 2305553, "postDate": "2023-06-16T18:59:15.540Z", "content": "

so in the end ICDAR is allowed and the competition was extended without any purpose?

", "rawMarkdown": "so in the end ICDAR is allowed and the competition was extended without any purpose?", "votes": 10 }, { "id": 2306334, "postDate": "2023-06-17T08:06:43.067Z", "content": "

These should have been the rules all along since it was recommended to us to train with ICDAR.
\nI believe it's best choice though, to avoid cleaning the leaderboard confusions after competition ends.

", "rawMarkdown": "These should have been the rules all along since it was recommended to us to train with ICDAR.\nI believe it's best choice though, to avoid cleaning the leaderboard confusions after competition ends.", "votes": 5, "replies": [ { "id": 2307997, "postDate": "2023-06-18T15:19:55.653Z", "content": "

Agreed, as that would be a lose lose for everyone. Just wish this all happened prior to a week before original competition end

", "rawMarkdown": "Agreed, as that would be a lose lose for everyone. Just wish this all happened prior to a week before original competition end", "votes": 3 } ] }, { "id": 2308677, "postDate": "2023-06-19T06:22:19.560Z", "content": "

great, updated 3 days ago, and we perfectly missed the information.

", "rawMarkdown": "great, updated 3 days ago, and we perfectly missed the information.", "votes": 6 }, { "id": 2305584, "postDate": "2023-06-16T19:27:41.517Z", "content": "

\n

Just to make sure, yolo v5 and yolo v8 license are not GPL 3.0. They are Gnu AGLPL 3.0. It would be good to state whether this license is also acceptable or not.

\n

Edited as I don't want to trigger more resentment. I didn't see it was 3 days before end of competition. I thought it was positive as it was undoing the insane situation where people are told to not use the dataset recommended by host.

", "rawMarkdown": "~~Great change!~~\n\nJust to make sure, yolo v5 and yolo v8 license are not GPL 3.0. They are Gnu AGLPL 3.0. It would be good to state whether this license is also acceptable or not.\n\nEdited as I don't want to trigger more resentment. I didn't see it was 3 days before end of competition. I thought it was positive as it was undoing the insane situation where people are told to not use the dataset recommended by host.", "votes": 4, "replies": [ { "id": 2305602, "postDate": "2023-06-16T19:42:55.953Z", "content": "

how is this a great change? 3 days before end?

", "rawMarkdown": "how is this a great change? 3 days before end?", "votes": 6, "replies": [ { "id": 2305606, "postDate": "2023-06-16T19:47:12.953Z", "content": "

Its a punch in the face for all who played fair and build a commercially usable solution. 3 days is not really a lot to look into data which now is newly allowed

", "rawMarkdown": "Its a punch in the face for all who played fair and build a commercially usable solution. 3 days is not really a lot to look into data which now is newly allowed", "votes": 8 }, { "id": 2305610, "postDate": "2023-06-16T19:51:53.623Z", "content": "

Same situation here.

", "rawMarkdown": "Same situation here.", "votes": 1 }, { "id": 2305735, "postDate": "2023-06-16T22:09:13.037Z", "content": "

I didn't see it was 3 days before end of competition. I thought it was positive as it was undoing the insane situation where people are told to not use the dataset recommended by host.

\n

I agree that a change like that only 3 days before end is bad.

", "rawMarkdown": "I didn't see it was 3 days before end of competition. I thought it was positive as it was undoing the insane situation where people are told to not use the dataset recommended by host.\n\nI agree that a change like that only 3 days before end is bad.", "votes": 4 }, { "id": 2306262, "postDate": "2023-06-17T07:34:30.573Z", "content": "

The main difference between the two is that:

\n\n

The lines [10-12] from the ultralytics license

\n
\n

The GNU Affero General Public License is a free, copyleft license for
\n software and other kinds of works, specifically designed to ensure
\n cooperation with the community in the case of network server software.

\n
", "rawMarkdown": "The main difference between the two is that: \n- GPL 3.0 (GNU General Public License version 3): If you use a piece of software that's licensed under the GPL, and you modify that software, you only need to share your changes if you give the software to someone else (like if you sell it or give it to a friend).\n\n- AGPL 3.0 (GNU Affero General Public License version 3): It's like the GPL, but with an extra rule. If you use a piece of software that's licensed under the AGPL, and you modify that software, you have to share your changes not only if you give the software to someone else, but also if you let people use it over a network (like if you put it on a server and let people use it as a web service).\n\nThe lines [10-12] from the [ultralytics license](https://github.com/ultralytics/ultralytics/blob/main/LICENSE)\n\n>The GNU Affero General Public License is a free, copyleft license for\nsoftware and other kinds of works, specifically designed to ensure\ncooperation with the community in the case of network server software." }, { "id": 2306287, "postDate": "2023-06-17T07:42:18.477Z", "content": "

Yes, Gnu AGPL is mor restrictive than GPL. It is why I asked.

", "rawMarkdown": "Yes, Gnu AGPL is mor restrictive than GPL. It is why I asked.", "votes": 1 }, { "id": 2306318, "postDate": "2023-06-17T07:56:58.490Z", "content": "

Only in the case of network server software.

", "rawMarkdown": "Only in the case of network server software." }, { "id": 2306350, "postDate": "2023-06-17T08:23:51.643Z", "content": "

Did I say something wrong?

", "rawMarkdown": "Did I say something wrong?" }, { "id": 2306357, "postDate": "2023-06-17T08:29:04.940Z", "content": "

Absolutely not, I confirmed your sentence is restrictive only in the case of network server software :)

", "rawMarkdown": "Absolutely not, I confirmed your sentence is restrictive only in the case of network server software :)" } ] }, { "id": 2305616, "postDate": "2023-06-16T19:55:43.423Z", "content": "

Another disappointment, the rules just should not have changed the first time and we should not have waited 2 weeks to get the final update.

", "rawMarkdown": "Another disappointment, the rules just should not have changed the first time and we should not have waited 2 weeks to get the final update. ", "votes": 3, "replies": [ { "id": 2305736, "postDate": "2023-06-16T22:09:48.800Z", "rawMarkdown": "", "isDeleted": true } ] }, { "id": 2305639, "postDate": "2023-06-16T20:09:54.230Z", "content": "

Changing rules three days before end of competition can never be a good change.

", "rawMarkdown": "Changing rules three days before end of competition can never be a good change. ", "votes": 11, "replies": [ { "id": 2305651, "postDate": "2023-06-16T20:17:04.323Z", "content": "

Not to mention this is now the second time we are doing this.

", "rawMarkdown": "Not to mention this is now the second time we are doing this. ", "votes": 3 } ] } ] }, { "id": 2306659, "postDate": "2023-06-17T13:11:44.260Z", "content": "

What?!!! I spent a week replace my yolov5 model to yolox……

", "rawMarkdown": "What?!!! I spent a week replace my yolov5 model to yolox......", "votes": 4, "replies": [ { "id": 2306683, "postDate": "2023-06-17T13:33:39.753Z", "content": "

I spent all night trying to incorporate the ICDAR dataset. Then it crashed at 7AM and it crashed again at noon 😨

", "rawMarkdown": "I spent all night trying to incorporate the ICDAR dataset. Then it crashed at 7AM and it crashed again at noon 😨", "replies": [ { "id": 2306691, "postDate": "2023-06-17T13:42:05.903Z", "rawMarkdown": "", "isDeleted": true }, { "id": 2306694, "postDate": "2023-06-17T13:47:18.320Z", "content": "

Now, most of things are engineering.

", "rawMarkdown": "Now, most of things are engineering." } ] }, { "id": 2306723, "postDate": "2023-06-17T14:06:11.757Z", "content": "

yolov5 is not allowed by this change unless mistaken.

", "rawMarkdown": "yolov5 is not allowed by this change unless mistaken.", "votes": 5, "replies": [ { "id": 2306736, "postDate": "2023-06-17T14:24:29.380Z", "content": "

I think CPMP is correct.

", "rawMarkdown": "I think CPMP is correct.", "votes": 2 } ] } ] }, { "id": 2305920, "postDate": "2023-06-17T02:11:53.417Z", "content": "

I am very interested to see if this dataset actually helps and how much.

", "rawMarkdown": "I am very interested to see if this dataset actually helps and how much.", "votes": 1, "replies": [ { "id": 2306348, "postDate": "2023-06-17T08:21:40.920Z", "content": "

I never use it at all. I use only the data published on the Kaggle and I achieve high performance on my local validation data.

", "rawMarkdown": "I never use it at all. I use only the data published on the Kaggle and I achieve high performance on my local validation data. " }, { "id": 2310001, "postDate": "2023-06-20T03:51:43.633Z", "rawMarkdown": "", "isDeleted": true } ] }, { "id": 2306883, "postDate": "2023-06-17T17:00:18.247Z", "content": "

What is \"in the spirit of fairness\" about a rule change 3 days ago?

", "rawMarkdown": "What is \"in the spirit of fairness\" about a rule change 3 days ago?", "votes": 2 }, { "id": 2305559, "postDate": "2023-06-16T19:05:15.483Z", "content": "

So now we have 3 days to try to re incorporate the ICDAR all over again? And as @alejopaullier points out we had the competition extension to come to the same result.

", "rawMarkdown": "So now we have 3 days to try to re incorporate the ICDAR all over again? And as @alejopaullier points out we had the competition extension to come to the same result. ", "votes": 2 }, { "id": 2305401, "postDate": "2023-06-16T17:10:48.260Z", "content": "

Thank you for reconsidering the license!
\nI hope you understand that we were frustrated by the ambiguous responses and silence with no apparent intent to respond, not by the availability of the license itself.
\nI also apologize for using slightly stronger language in some threads.

", "rawMarkdown": "Thank you for reconsidering the license!\nI hope you understand that we were frustrated by the ambiguous responses and silence with no apparent intent to respond, not by the availability of the license itself.\nI also apologize for using slightly stronger language in some threads.", "votes": 1 }, { "id": 2305652, "postDate": "2023-06-16T20:18:25.133Z", "content": "

@rkeiii
\nHow about extending the competition another 4 days? The host banned the ICDAR dataset, which had been recommended or allowed since the beginning of the competition, 7 days before the original deadline, and now re-allowed it 3 days before the extended deadline. Participants probably lost 4 days (plus the extended 7 days) in this mess.

", "rawMarkdown": "@rkeiii \nHow about extending the competition another 4 days? The host banned the ICDAR dataset, which had been recommended or allowed since the beginning of the competition, 7 days before the original deadline, and now re-allowed it 3 days before the extended deadline. Participants probably lost 4 days (plus the extended 7 days) in this mess.", "votes": -3, "replies": [ { "id": 2305656, "postDate": "2023-06-16T20:22:20.853Z", "content": "

At this point what is done is done. Rule changes and continuous extensions have drawn out this competition. Personally I would rather put this one behind us.

", "rawMarkdown": "At this point what is done is done. Rule changes and continuous extensions have drawn out this competition. Personally I would rather put this one behind us.", "votes": 11, "replies": [ { "id": 2305664, "postDate": "2023-06-16T20:33:27.237Z", "content": "

Yes, I honestly think so too. It will probably take a few more days to decide on an extension, so if no one feels unfair about the 4 days lost, I do not want any more extensions.

", "rawMarkdown": "Yes, I honestly think so too. It will probably take a few more days to decide on an extension, so if no one feels unfair about the 4 days lost, I do not want any more extensions." } ] }, { "id": 2305665, "postDate": "2023-06-16T20:33:59.360Z", "content": "

Our team also spent the whole last week trying to get back to our best score without ICDAR, but I don't think a second extension is a good idea. At this point let's just get it over with.

", "rawMarkdown": "Our team also spent the whole last week trying to get back to our best score without ICDAR, but I don't think a second extension is a good idea. At this point let's just get it over with.", "votes": 15, "replies": [ { "id": 2305667, "postDate": "2023-06-16T20:37:57.870Z", "content": "

I agree with you. I don't want to spend any more time on this competition. Let's end it. (editied)

", "rawMarkdown": "I agree with you. I don't want to spend any more time on this competition. Let's end it. (editied)", "votes": 3 }, { "id": 2305668, "postDate": "2023-06-16T20:38:05.120Z", "content": "

Completely agree. Same here, we spent more than one week to get rid of ICDAR but such is life, let's finish this and move on!

", "rawMarkdown": "Completely agree. Same here, we spent more than one week to get rid of ICDAR but such is life, let's finish this and move on!", "votes": 10 }, { "id": 2305887, "postDate": "2023-06-17T01:25:32.883Z", "content": "

I agree with you.

\n

The longer this issue goes on, the more irreversible it becomes, and I don't think there is any way to round it up now that it has been two weeks since the initial statement was made.

\n

If the host continues to refuse to allow the use of the dataset, the efforts of the participants who worked diligently on it since the early days will be wasted, and if they allow the use of the dataset now, the efforts of all the participants in the last week will be wasted.
\nI myself spent the first few months trying to figure out how to use the ICDAR dataset, and the last week trying to figure out how to improve accuracy without using the ICDAR dataset, so the damage is done either way.

\n

But personally, I'd hate to see the hosts change their statements around any more and confuse us or re-extend the time period and make it an endless competition.
\nThere is probably no way for everyone to benefit from the use of the data set, whether they allow it or not, so If the hosts decide to allow it, that's fine, but above all, I would like to see the competition end properly at the current date.

", "rawMarkdown": "I agree with you.\n \nThe longer this issue goes on, the more irreversible it becomes, and I don't think there is any way to round it up now that it has been two weeks since the initial statement was made.\n\nIf the host continues to refuse to allow the use of the dataset, the efforts of the participants who worked diligently on it since the early days will be wasted, and if they allow the use of the dataset now, the efforts of all the participants in the last week will be wasted.\nI myself spent the first few months trying to figure out how to use the ICDAR dataset, and the last week trying to figure out how to improve accuracy without using the ICDAR dataset, so the damage is done either way.\n\nBut personally, I'd hate to see the hosts change their statements around any more and confuse us or re-extend the time period and make it an endless competition.\nThere is probably no way for everyone to benefit from the use of the data set, whether they allow it or not, so If the hosts decide to allow it, that's fine, but above all, I would like to see the competition end properly at the current date.", "votes": 6 } ] } ] }, { "id": 2305389, "postDate": "2023-06-16T16:57:05.623Z", "content": "

I want to apologize for any confusion participants may have experienced as a result of rule changes that occurred.

\n

Out of respect for the hard work all of the participants have invested in their entries and in the spirit of fairness, we will NOT be disqualifying or otherwise penalizing participants due to use of datasets or pre-trained models licensed under any of the following:

\n\n

Thank you again to all of our participants for your hard work on this difficult problem. Please know we have been working diligently behind the scenes to ensure your valuable time and effort working with this data will not go to waste.

", "rawMarkdown": "I want to apologize for any confusion participants may have experienced as a result of rule changes that occurred.\n\nOut of respect for the hard work all of the participants have invested in their entries and in the spirit of fairness, we will NOT be disqualifying or otherwise penalizing participants due to use of datasets or pre-trained models licensed under any of the following:\n- MIT\n- CC NC-BY-SA 4.0\n- GPL-3.0\n\nThank you again to all of our participants for your hard work on this difficult problem. Please know we have been working diligently behind the scenes to ensure your valuable time and effort working with this data will not go to waste.", "votes": -27 }, { "id": 2306253, "postDate": "2023-06-17T07:20:52.310Z", "content": "

You forgot the more permissive and attractive license than GPL 3.0 which is Apache 2.0. Commercial companies often prefer Apache 2.0 over GPL 3.0 due to its permissiveness, allowing proprietary use and modifications. Apache 2.0's explicit patent rights grant provides added legal protection. It is compatible with many licenses and favored for its legal clarity, especially around contributions and patent rights. Could you please add it to the list?

", "rawMarkdown": "**You forgot the more permissive and attractive license than GPL 3.0 which is Apache 2.0.** Commercial companies often prefer Apache 2.0 over GPL 3.0 due to its permissiveness, allowing proprietary use and modifications. Apache 2.0's explicit patent rights grant provides added legal protection. It is compatible with many licenses and favored for its legal clarity, especially around contributions and patent rights. Could you please add it to the list?", "replies": [ { "id": 2306276, "postDate": "2023-06-17T07:38:39.800Z", "content": "

Apache 2.0 allow commercial use and is compatible with the rules. What they change with the above announce is that some licenses that were incompatible with the rules are now accepted.

", "rawMarkdown": "Apache 2.0 allow commercial use and is compatible with the rules. What they change with the above announce is that some licenses that were incompatible with the rules are now accepted.", "votes": 4 } ] }, { "id": 2306036, "postDate": "2023-06-17T04:31:32.503Z", "content": "

@rkeiii

\n
\n

Just to make sure, yolo v5 and yolo v8 license are not GPL 3.0. They are Gnu AGLPL 3.0. It would be good to state whether this license is also acceptable or not.

\n
", "rawMarkdown": "@rkeiii \n\n> Just to make sure, yolo v5 and yolo v8 license are not GPL 3.0. They are Gnu AGLPL 3.0. It would be good to state whether this license is also acceptable or not.\n\n", "replies": [ { "id": 2306148, "postDate": "2023-06-17T06:01:26.157Z", "content": "

I understand that CC NC-BY-SA 4.0 is allowed, but is CC BY-NC-ND 4.0 also allowed?

", "rawMarkdown": "I understand that CC NC-BY-SA 4.0 is allowed, but is CC BY-NC-ND 4.0 also allowed?", "replies": [ { "id": 2306231, "postDate": "2023-06-17T06:57:04.727Z", "content": "

For now, I believe CC BY-NC-ND is not available. Also, for the same reason, AGPL 3.0 is not available.

", "rawMarkdown": "For now, I believe CC BY-NC-ND is not available. Also, for the same reason, AGPL 3.0 is not available.", "votes": 2 }, { "id": 2310201, "postDate": "2023-06-20T07:38:48.840Z", "content": "

CC BY-NC-ND 4.0 was related to adobe synthetic dataset? Had a lot of scatter plots

", "rawMarkdown": "CC BY-NC-ND 4.0 was related to adobe synthetic dataset? Had a lot of scatter plots" }, { "id": 2310215, "postDate": "2023-06-20T07:55:43.930Z", "content": "

That's exactly right. I prepared for it, but never got a reply, so I didn't end up using it.

", "rawMarkdown": "That's exactly right. I prepared for it, but never got a reply, so I didn't end up using it.", "votes": 1 } ] } ] }, { "id": 2306275, "postDate": "2023-06-17T07:38:11.643Z", "rawMarkdown": "", "isDeleted": true }, { "id": 2305624, "postDate": "2023-06-16T20:01:18.993Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 2305601, "author_name": "Dieter", "author_url": "", "post_date": "2023-06-16T19:42:21.063000", "content": "

A punch in the face for all that played fair…

", "votes": 14, "replies": [ { "id": 2305618, "author_name": "CroDoc", "author_url": "", "post_date": "2023-06-16T19:56:45.987000", "content": "

We have a full MIT compliant solution. We spent countless days to make it work after the latest rules change.

", "votes": 8, "replies": [ { "id": 2305621, "author_name": "Cody_Null", "author_url": "", "post_date": "2023-06-16T19:58:28.853000", "content": "

We did the same, despite it not being the easiest or the most effective method.

", "votes": 3, "replies": [] } ] }, { "id": 2306057, "author_name": "Thanh Hau Nguyen", "author_url": "", "post_date": "2023-06-17T04:46:00.567000", "content": "

Definitely Agree!

", "votes": 0, "replies": [] } ] }, { "id": 2306826, "author_name": "Nicholas Broad", "author_url": "", "post_date": "2023-06-17T16:05:37.400000", "content": "

@maggiemd,

\n

I think it would be great if all future competition organizers explicitly said which licenses were acceptable at the beginning of the competition so we can avoid situations like this.

", "votes": 10, "replies": [ { "id": 2306950, "author_name": "CPMP", "author_url": "", "post_date": "2023-06-17T17:44:56.577000", "content": "

It was stated clearly in the rules. What was insane is that host recommended a dataset incompatible with the rules.

", "votes": 3, "replies": [ { "id": 2307016, "author_name": "Nicholas Broad", "author_url": "", "post_date": "2023-06-17T18:32:06.890000", "content": "

The rules say

\n
\n
    \n
  1. SUBMISSION CODE REQUIREMENTS. Use of Open Source
    \n The following supersedes Section 8.C of the rules below.
    \n Use of Open Source. If open source code is used in the model to generate the Submission, then you must only use open source code licensed under an Open Source Initiative-approved license (see www.opensource.org) that in no event limits commercial use of such code or model containing or depending on such code. Reference was made to an outside data set, as an example; however, we have determined that example does not meet Sponsor’s open source requirements. For clarification, we have removed the reference to that outside data set. If you wish to use any training data set other than those provided on the Competition Website, it is your responsibility to verify that, given the possibility of financial remuneration, you have the necessary rights to use such other data set.
    2. \n
\n
\n
\n
\n

7C. External Data. You may use data other than the Competition Data (“External Data”) to develop and test your Submissions. However, you will ensure the External Data is publicly available and equally accessible to use by all participants of the Competition for purposes of the competition at no cost to the other participants. The ability to use External Data under this Section 7.C (External Data) does not limit your other obligations under these Competition Rules, including but not limited to Section 11 (Winners Obligations).

\n
\n
\n
\n

8C. Use of Open Source. Unless otherwise stated in the Specific Competition Rules above, if open source code is used in the model to generate the Submission, then you must only use open source code licensed under an Open Source Initiative-approved license (see www.opensource.org) that in no event limits commercial use of such code or model containing or depending on such code.

\n
\n

The AGPL-3.0 license is listed on opensource.org, so I think there could still be confusion without the hosts explicitly indicating allowed licenses

", "votes": 0, "replies": [] }, { "id": 2307055, "author_name": "CPMP", "author_url": "", "post_date": "2023-06-17T19:18:36.867000", "content": "
\n

that in no event limits commercial use of such code or model containing or depending on such code.

\n
\n

Are you saying GNU AGPL 3.0 meets this?

", "votes": 0, "replies": [] }, { "id": 2307070, "author_name": "Nicholas Broad", "author_url": "", "post_date": "2023-06-17T19:34:53.833000", "content": "

I'm not a legal expert, but it seems ok to me

\n

https://www.tldrlegal.com/license/gnu-affero-general-public-license-v3-agpl-3-0

", "votes": 0, "replies": [] } ] } ] }, { "id": 2305591, "author_name": "CroDoc", "author_url": "", "post_date": "2023-06-16T19:32:50.090000", "content": "

After spending countless additional days to make our code MIT compliant, this decision (3 days before the end) comes as a complete shock to me. Is this a joke? Again?

", "votes": 10, "replies": [ { "id": 2305592, "author_name": "Benedikt Droste", "author_url": "", "post_date": "2023-06-16T19:34:22.143000", "content": "

Same here, unbelievable…

", "votes": 3, "replies": [] } ] }, { "id": 2305553, "author_name": "moth", "author_url": "", "post_date": "2023-06-16T18:59:15.540000", "content": "

so in the end ICDAR is allowed and the competition was extended without any purpose?

", "votes": 10, "replies": [] }, { "id": 2306334, "author_name": "Rafi Hai", "author_url": "", "post_date": "2023-06-17T08:06:43.067000", "content": "

These should have been the rules all along since it was recommended to us to train with ICDAR.
\nI believe it's best choice though, to avoid cleaning the leaderboard confusions after competition ends.

", "votes": 5, "replies": [ { "id": 2307997, "author_name": "Cody_Null", "author_url": "", "post_date": "2023-06-18T15:19:55.653000", "content": "

Agreed, as that would be a lose lose for everyone. Just wish this all happened prior to a week before original competition end

", "votes": 3, "replies": [] } ] }, { "id": 2308677, "author_name": "Mengfei Li", "author_url": "", "post_date": "2023-06-19T06:22:19.560000", "content": "

great, updated 3 days ago, and we perfectly missed the information.

", "votes": 6, "replies": [] }, { "id": 2305584, "author_name": "CPMP", "author_url": "", "post_date": "2023-06-16T19:27:41.517000", "content": "

\n

Just to make sure, yolo v5 and yolo v8 license are not GPL 3.0. They are Gnu AGLPL 3.0. It would be good to state whether this license is also acceptable or not.

\n

Edited as I don't want to trigger more resentment. I didn't see it was 3 days before end of competition. I thought it was positive as it was undoing the insane situation where people are told to not use the dataset recommended by host.

", "votes": 4, "replies": [ { "id": 2305602, "author_name": "Dieter", "author_url": "", "post_date": "2023-06-16T19:42:55.953000", "content": "

how is this a great change? 3 days before end?

", "votes": 6, "replies": [ { "id": 2305606, "author_name": "Dieter", "author_url": "", "post_date": "2023-06-16T19:47:12.953000", "content": "

Its a punch in the face for all who played fair and build a commercially usable solution. 3 days is not really a lot to look into data which now is newly allowed

", "votes": 8, "replies": [] }, { "id": 2305610, "author_name": "CroDoc", "author_url": "", "post_date": "2023-06-16T19:51:53.623000", "content": "

Same situation here.

", "votes": 1, "replies": [] }, { "id": 2305735, "author_name": "CPMP", "author_url": "", "post_date": "2023-06-16T22:09:13.037000", "content": "

I didn't see it was 3 days before end of competition. I thought it was positive as it was undoing the insane situation where people are told to not use the dataset recommended by host.

\n

I agree that a change like that only 3 days before end is bad.

", "votes": 4, "replies": [] }, { "id": 2306262, "author_name": "Hamzah", "author_url": "", "post_date": "2023-06-17T07:34:30.573000", "content": "

The main difference between the two is that:

\n\n

The lines [10-12] from the ultralytics license

\n
\n

The GNU Affero General Public License is a free, copyleft license for
\n software and other kinds of works, specifically designed to ensure
\n cooperation with the community in the case of network server software.

\n
", "votes": 0, "replies": [] }, { "id": 2306287, "author_name": "CPMP", "author_url": "", "post_date": "2023-06-17T07:42:18.477000", "content": "

Yes, Gnu AGPL is mor restrictive than GPL. It is why I asked.

", "votes": 1, "replies": [] }, { "id": 2306318, "author_name": "Hamzah", "author_url": "", "post_date": "2023-06-17T07:56:58.490000", "content": "

Only in the case of network server software.

", "votes": 0, "replies": [] }, { "id": 2306350, "author_name": "CPMP", "author_url": "", "post_date": "2023-06-17T08:23:51.643000", "content": "

Did I say something wrong?

", "votes": 0, "replies": [] }, { "id": 2306357, "author_name": "Hamzah", "author_url": "", "post_date": "2023-06-17T08:29:04.940000", "content": "

Absolutely not, I confirmed your sentence is restrictive only in the case of network server software :)

", "votes": 0, "replies": [] } ] }, { "id": 2305616, "author_name": "Cody_Null", "author_url": "", "post_date": "2023-06-16T19:55:43.423000", "content": "

Another disappointment, the rules just should not have changed the first time and we should not have waited 2 weeks to get the final update.

", "votes": 3, "replies": [ { "id": 2305736, "author_name": "", "author_url": "", "post_date": "2023-06-16T22:09:48.800000", "content": "", "votes": 0, "replies": [] } ] }, { "id": 2305639, "author_name": "Psi", "author_url": "", "post_date": "2023-06-16T20:09:54.230000", "content": "

Changing rules three days before end of competition can never be a good change.

", "votes": 11, "replies": [ { "id": 2305651, "author_name": "Cody_Null", "author_url": "", "post_date": "2023-06-16T20:17:04.323000", "content": "

Not to mention this is now the second time we are doing this.

", "votes": 3, "replies": [] } ] } ] }, { "id": 2306659, "author_name": "sheep", "author_url": "", "post_date": "2023-06-17T13:11:44.260000", "content": "

What?!!! I spent a week replace my yolov5 model to yolox……

", "votes": 4, "replies": [ { "id": 2306683, "author_name": "CroDoc", "author_url": "", "post_date": "2023-06-17T13:33:39.753000", "content": "

I spent all night trying to incorporate the ICDAR dataset. Then it crashed at 7AM and it crashed again at noon 😨

", "votes": 0, "replies": [ { "id": 2306691, "author_name": "", "author_url": "", "post_date": "2023-06-17T13:42:05.903000", "content": "", "votes": 0, "replies": [] }, { "id": 2306694, "author_name": "sheep", "author_url": "", "post_date": "2023-06-17T13:47:18.320000", "content": "

Now, most of things are engineering.

", "votes": 0, "replies": [] } ] }, { "id": 2306723, "author_name": "CPMP", "author_url": "", "post_date": "2023-06-17T14:06:11.757000", "content": "

yolov5 is not allowed by this change unless mistaken.

", "votes": 5, "replies": [ { "id": 2306736, "author_name": "CroDoc", "author_url": "", "post_date": "2023-06-17T14:24:29.380000", "content": "

I think CPMP is correct.

", "votes": 2, "replies": [] } ] } ] }, { "id": 2305920, "author_name": "CroDoc", "author_url": "", "post_date": "2023-06-17T02:11:53.417000", "content": "

I am very interested to see if this dataset actually helps and how much.

", "votes": 1, "replies": [ { "id": 2306348, "author_name": "Hamzah", "author_url": "", "post_date": "2023-06-17T08:21:40.920000", "content": "

I never use it at all. I use only the data published on the Kaggle and I achieve high performance on my local validation data.

", "votes": 0, "replies": [] }, { "id": 2310001, "author_name": "", "author_url": "", "post_date": "2023-06-20T03:51:43.633000", "content": "", "votes": 0, "replies": [] } ] }, { "id": 2306883, "author_name": "Isao", "author_url": "", "post_date": "2023-06-17T17:00:18.247000", "content": "

What is \"in the spirit of fairness\" about a rule change 3 days ago?

", "votes": 2, "replies": [] }, { "id": 2305559, "author_name": "Cody_Null", "author_url": "", "post_date": "2023-06-16T19:05:15.483000", "content": "

So now we have 3 days to try to re incorporate the ICDAR all over again? And as @alejopaullier points out we had the competition extension to come to the same result.

", "votes": 2, "replies": [] }, { "id": 2305401, "author_name": "penguin46", "author_url": "", "post_date": "2023-06-16T17:10:48.260000", "content": "

Thank you for reconsidering the license!
\nI hope you understand that we were frustrated by the ambiguous responses and silence with no apparent intent to respond, not by the availability of the license itself.
\nI also apologize for using slightly stronger language in some threads.

", "votes": 1, "replies": [] }, { "id": 2305652, "author_name": "penguin46", "author_url": "", "post_date": "2023-06-16T20:18:25.133000", "content": "

@rkeiii
\nHow about extending the competition another 4 days? The host banned the ICDAR dataset, which had been recommended or allowed since the beginning of the competition, 7 days before the original deadline, and now re-allowed it 3 days before the extended deadline. Participants probably lost 4 days (plus the extended 7 days) in this mess.

", "votes": -3, "replies": [ { "id": 2305656, "author_name": "Cody_Null", "author_url": "", "post_date": "2023-06-16T20:22:20.853000", "content": "

At this point what is done is done. Rule changes and continuous extensions have drawn out this competition. Personally I would rather put this one behind us.

", "votes": 11, "replies": [ { "id": 2305664, "author_name": "penguin46", "author_url": "", "post_date": "2023-06-16T20:33:27.237000", "content": "

Yes, I honestly think so too. It will probably take a few more days to decide on an extension, so if no one feels unfair about the 4 days lost, I do not want any more extensions.

", "votes": 0, "replies": [] } ] }, { "id": 2305665, "author_name": "qdv206", "author_url": "", "post_date": "2023-06-16T20:33:59.360000", "content": "

Our team also spent the whole last week trying to get back to our best score without ICDAR, but I don't think a second extension is a good idea. At this point let's just get it over with.

", "votes": 15, "replies": [ { "id": 2305667, "author_name": "penguin46", "author_url": "", "post_date": "2023-06-16T20:37:57.870000", "content": "

I agree with you. I don't want to spend any more time on this competition. Let's end it. (editied)

", "votes": 3, "replies": [] }, { "id": 2305668, "author_name": "Eduardo Rocha de Andrade", "author_url": "", "post_date": "2023-06-16T20:38:05.120000", "content": "

Completely agree. Same here, we spent more than one week to get rid of ICDAR but such is life, let's finish this and move on!

", "votes": 10, "replies": [] }, { "id": 2305887, "author_name": "YumeNeko", "author_url": "", "post_date": "2023-06-17T01:25:32.883000", "content": "

I agree with you.

\n

The longer this issue goes on, the more irreversible it becomes, and I don't think there is any way to round it up now that it has been two weeks since the initial statement was made.

\n

If the host continues to refuse to allow the use of the dataset, the efforts of the participants who worked diligently on it since the early days will be wasted, and if they allow the use of the dataset now, the efforts of all the participants in the last week will be wasted.
\nI myself spent the first few months trying to figure out how to use the ICDAR dataset, and the last week trying to figure out how to improve accuracy without using the ICDAR dataset, so the damage is done either way.

\n

But personally, I'd hate to see the hosts change their statements around any more and confuse us or re-extend the time period and make it an endless competition.
\nThere is probably no way for everyone to benefit from the use of the data set, whether they allow it or not, so If the hosts decide to allow it, that's fine, but above all, I would like to see the competition end properly at the current date.

", "votes": 6, "replies": [] } ] } ] }, { "id": 2306253, "author_name": "Hamzah", "author_url": "", "post_date": "2023-06-17T07:20:52.310000", "content": "

You forgot the more permissive and attractive license than GPL 3.0 which is Apache 2.0. Commercial companies often prefer Apache 2.0 over GPL 3.0 due to its permissiveness, allowing proprietary use and modifications. Apache 2.0's explicit patent rights grant provides added legal protection. It is compatible with many licenses and favored for its legal clarity, especially around contributions and patent rights. Could you please add it to the list?

", "votes": 0, "replies": [ { "id": 2306276, "author_name": "CPMP", "author_url": "", "post_date": "2023-06-17T07:38:39.800000", "content": "

Apache 2.0 allow commercial use and is compatible with the rules. What they change with the above announce is that some licenses that were incompatible with the rules are now accepted.

", "votes": 4, "replies": [] } ] }, { "id": 2306036, "author_name": "YujiAriyasu", "author_url": "", "post_date": "2023-06-17T04:31:32.503000", "content": "

@rkeiii

\n
\n

Just to make sure, yolo v5 and yolo v8 license are not GPL 3.0. They are Gnu AGLPL 3.0. It would be good to state whether this license is also acceptable or not.

\n
", "votes": 0, "replies": [ { "id": 2306148, "author_name": "YujiAriyasu", "author_url": "", "post_date": "2023-06-17T06:01:26.157000", "content": "

I understand that CC NC-BY-SA 4.0 is allowed, but is CC BY-NC-ND 4.0 also allowed?

", "votes": 0, "replies": [ { "id": 2306231, "author_name": "YujiAriyasu", "author_url": "", "post_date": "2023-06-17T06:57:04.727000", "content": "

For now, I believe CC BY-NC-ND is not available. Also, for the same reason, AGPL 3.0 is not available.

", "votes": 2, "replies": [] }, { "id": 2310201, "author_name": "Dieter", "author_url": "", "post_date": "2023-06-20T07:38:48.840000", "content": "

CC BY-NC-ND 4.0 was related to adobe synthetic dataset? Had a lot of scatter plots

", "votes": 0, "replies": [] }, { "id": 2310215, "author_name": "YujiAriyasu", "author_url": "", "post_date": "2023-06-20T07:55:43.930000", "content": "

That's exactly right. I prepared for it, but never got a reply, so I didn't end up using it.

", "votes": 1, "replies": [] } ] } ] }, { "id": 2306275, "author_name": "", "author_url": "", "post_date": "2023-06-17T07:38:11.643000", "content": "", "votes": 0, "replies": [] }, { "id": 2305624, "author_name": "", "author_url": "", "post_date": "2023-06-16T20:01:18.993000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "2305601": "A punch in the face for all that played fair...", "2306826": "@maggiemd,\n\nI think it would be great if all future competition organizers explicitly said which licenses were acceptable at the beginning of the competition so we can avoid situations like this. ", "2305591": "After spending countless additional days to make our code MIT compliant, this decision (3 days before the end) comes as a complete shock to me. Is this a joke? Again?", "2305553": "so in the end ICDAR is allowed and the competition was extended without any purpose?", "2306334": "These should have been the rules all along since it was recommended to us to train with ICDAR.\nI believe it's best choice though, to avoid cleaning the leaderboard confusions after competition ends.", "2308677": "great, updated 3 days ago, and we perfectly missed the information.", "2305584": "~~Great change!~~\n\nJust to make sure, yolo v5 and yolo v8 license are not GPL 3.0. They are Gnu AGLPL 3.0. It would be good to state whether this license is also acceptable or not.\n\nEdited as I don't want to trigger more resentment. I didn't see it was 3 days before end of competition. I thought it was positive as it was undoing the insane situation where people are told to not use the dataset recommended by host.", "2306659": "What?!!! I spent a week replace my yolov5 model to yolox......", "2305920": "I am very interested to see if this dataset actually helps and how much.", "2306883": "What is \"in the spirit of fairness\" about a rule change 3 days ago?", "2305559": "So now we have 3 days to try to re incorporate the ICDAR all over again? And as @alejopaullier points out we had the competition extension to come to the same result. ", "2305401": "Thank you for reconsidering the license!\nI hope you understand that we were frustrated by the ambiguous responses and silence with no apparent intent to respond, not by the availability of the license itself.\nI also apologize for using slightly stronger language in some threads.", "2305652": "@rkeiii \nHow about extending the competition another 4 days? The host banned the ICDAR dataset, which had been recommended or allowed since the beginning of the competition, 7 days before the original deadline, and now re-allowed it 3 days before the extended deadline. Participants probably lost 4 days (plus the extended 7 days) in this mess.", "2305389": "I want to apologize for any confusion participants may have experienced as a result of rule changes that occurred.\n\nOut of respect for the hard work all of the participants have invested in their entries and in the spirit of fairness, we will NOT be disqualifying or otherwise penalizing participants due to use of datasets or pre-trained models licensed under any of the following:\n- MIT\n- CC NC-BY-SA 4.0\n- GPL-3.0\n\nThank you again to all of our participants for your hard work on this difficult problem. Please know we have been working diligently behind the scenes to ensure your valuable time and effort working with this data will not go to waste.", "2306253": "**You forgot the more permissive and attractive license than GPL 3.0 which is Apache 2.0.** Commercial companies often prefer Apache 2.0 over GPL 3.0 due to its permissiveness, allowing proprietary use and modifications. Apache 2.0's explicit patent rights grant provides added legal protection. It is compatible with many licenses and favored for its legal clarity, especially around contributions and patent rights. Could you please add it to the list?", "2306036": "@rkeiii \n\n> Just to make sure, yolo v5 and yolo v8 license are not GPL 3.0. They are Gnu AGLPL 3.0. It would be good to state whether this license is also acceptable or not.\n\n", "2306275": "", "2305624": "" } }