Image IDs of Test set images

Question

Image IDs of Test set images

YeahMao opened this issue 4 years ago · 6 comments

Hi PubLayNet Team,

Thank you for sharing the dataset. I am wondering is there a Json file (or a list of image id) of the Test set provided?

Answer 1 · 2020-11-04T02:07:57.000Z

Hi YehMao,

Currently, there is no public ground truth for the test set. The ground truth is being reserved for the following ICDAR competition:

https://icdar2021.org/competitions/competition-on-scientific-literature-parsing

Answer 2 · 2020-11-04T04:30:05.000Z

Hi YehMao,

Currently, there is no public ground truth for the test set. The ground truth is being reserved for the following ICDAR competition:

https://icdar2021.org/competitions/competition-on-scientific-literature-parsing

Hi ajjimeno,

Thank you for your reply.

Yes, I participated in this competition, and I am trying to make a submission. I can't pass the phase if I don't have the correct image IDs. Therefore, I am wondering how can I get the corresponding id of an image. For train and val, I saw the IDs by checking the "train.json" and "val.json" files. For the test set, is there an alternative way to find the image IDs?

Thanks.

Answer 3 · 2020-11-05T04:22:36.000Z

Hi YeahMao, Thank you for pointing it out. We are preparing the file linking the file names and ids and will release it soon.

…

On Wed, Nov 4, 2020 at 3:30 PM YeahMao ***@***.***> wrote: Hi YehMao, Currently, there is no public ground truth for the test set. The ground truth is being reserved for the following ICDAR competition: https://icdar2021.org/competitions/competition-on-scientific-literature-parsing Hi ajjimeno, Thank you for your reply. Yes, I participated in this competition, and I am trying to make a submission. I can't pass the phase if I don't have the correct image IDs. Therefore, I am wondering how can I get the corresponding id of an image. For train and val, I saw the IDs by checking the "train.json" and "val.json" files. For the test set, is there an alternative way to find the image IDs? Thanks. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#29 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AA6BZDNRW3FY3NXQCB4KPR3SODKFVANCNFSM4THGGZMA> .

Answer 4 · 2020-11-08T09:25:42.000Z

I have made available the ids from the github of the challenge: https://github.com/ibm-aur-nlp/PubLayNet/blob/master/ICDAR_SLR_competition/test_ids.json Best regards, Antonio On Thu, Nov 5, 2020 at 3:22 PM Antonio Jimeno <antonio.jimeno@gmail.com> wrote:

…

Hi YeahMao, Thank you for pointing it out. We are preparing the file linking the file names and ids and will release it soon. On Wed, Nov 4, 2020 at 3:30 PM YeahMao ***@***.***> wrote: > Hi YehMao, > > Currently, there is no public ground truth for the test set. The ground > truth is being reserved for the following ICDAR competition: > > > https://icdar2021.org/competitions/competition-on-scientific-literature-parsing > > Hi ajjimeno, > > Thank you for your reply. > > Yes, I participated in this competition, and I am trying to make a > submission. I can't pass the phase if I don't have the correct image IDs. > Therefore, I am wondering how can I get the corresponding id of an image. > For train and val, I saw the IDs by checking the "train.json" and > "val.json" files. For the test set, is there an alternative way to find the > image IDs? > > Thanks. > > — > You are receiving this because you commented. > Reply to this email directly, view it on GitHub > <#29 (comment)>, > or unsubscribe > <https://github.com/notifications/unsubscribe-auth/AA6BZDNRW3FY3NXQCB4KPR3SODKFVANCNFSM4THGGZMA> > . >

Answer 5 · 2020-11-09T07:26:13.000Z

Hi Antonio, Thank you for the update. I have another question regarding the ICDAR competition. Is there a ground truth for the mini validation set (20 examples)? I noticed that there is a "anntations.png" file provided which is converted from the "samples.json" file. On the competition page, there is a host submission that matches this Json file. Since the overall score of it is not one out of one, I reckon it is just a baseline? Or should we regard the annotations.png as a ground truth? Thank you in advance. Best regards, Tony

…

On Sun, 8 Nov 2020 at 20:25, ajjimeno ***@***.***> wrote: I have made available the ids from the github of the challenge: https://github.com/ibm-aur-nlp/PubLayNet/blob/master/ICDAR_SLR_competition/test_ids.json Best regards, Antonio On Thu, Nov 5, 2020 at 3:22 PM Antonio Jimeno ***@***.***> wrote: > > Hi YeahMao, > > Thank you for pointing it out. We are preparing the file linking the > file names and ids and will release it soon. > > On Wed, Nov 4, 2020 at 3:30 PM YeahMao ***@***.***> wrote: > >> Hi YehMao, >> >> Currently, there is no public ground truth for the test set. The ground >> truth is being reserved for the following ICDAR competition: >> >> >> https://icdar2021.org/competitions/competition-on-scientific-literature-parsing >> >> Hi ajjimeno, >> >> Thank you for your reply. >> >> Yes, I participated in this competition, and I am trying to make a >> submission. I can't pass the phase if I don't have the correct image IDs. >> Therefore, I am wondering how can I get the corresponding id of an image. >> For train and val, I saw the IDs by checking the "train.json" and >> "val.json" files. For the test set, is there an alternative way to find the >> image IDs? >> >> Thanks. >> >> — >> You are receiving this because you commented. >> Reply to this email directly, view it on GitHub >> < #29 (comment) >, >> or unsubscribe >> < https://github.com/notifications/unsubscribe-auth/AA6BZDNRW3FY3NXQCB4KPR3SODKFVANCNFSM4THGGZMA > >> . >> > — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#29 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AQ5AKFNZGKV2FV26Z6D5TMTSOZP2DANCNFSM4THGGZMA> .

Answer 6 · 2020-11-09T23:30:20.000Z

Hi Tony, The annotations for those files should be available from the development/validation set. Please let me know if you have any issue finding them. Best regards, Antonio

…

On Mon, Nov 9, 2020 at 6:26 PM YeahMao ***@***.***> wrote: Hi Antonio, Thank you for the update. I have another question regarding the ICDAR competition. Is there a ground truth for the mini validation set (20 examples)? I noticed that there is a "anntations.png" file provided which is converted from the "samples.json" file. On the competition page, there is a host submission that matches this Json file. Since the overall score of it is not one out of one, I reckon it is just a baseline? Or should we regard the annotations.png as a ground truth? Thank you in advance. Best regards, Tony On Sun, 8 Nov 2020 at 20:25, ajjimeno ***@***.***> wrote: > I have made available the ids from the github of the challenge: > > https://github.com/ibm-aur-nlp/PubLayNet/blob/master/ICDAR_SLR_competition/test_ids.json > > Best regards, > Antonio > > On Thu, Nov 5, 2020 at 3:22 PM Antonio Jimeno ***@***.***> > wrote: > > > > > Hi YeahMao, > > > > Thank you for pointing it out. We are preparing the file linking the > > file names and ids and will release it soon. > > > > On Wed, Nov 4, 2020 at 3:30 PM YeahMao ***@***.***> wrote: > > > >> Hi YehMao, > >> > >> Currently, there is no public ground truth for the test set. The ground > >> truth is being reserved for the following ICDAR competition: > >> > >> > >> > https://icdar2021.org/competitions/competition-on-scientific-literature-parsing > >> > >> Hi ajjimeno, > >> > >> Thank you for your reply. > >> > >> Yes, I participated in this competition, and I am trying to make a > >> submission. I can't pass the phase if I don't have the correct image > IDs. > >> Therefore, I am wondering how can I get the corresponding id of an > image. > >> For train and val, I saw the IDs by checking the "train.json" and > >> "val.json" files. For the test set, is there an alternative way to find > the > >> image IDs? > >> > >> Thanks. > >> > >> — > >> You are receiving this because you commented. > >> Reply to this email directly, view it on GitHub > >> < > #29 (comment) > >, > >> or unsubscribe > >> < > https://github.com/notifications/unsubscribe-auth/AA6BZDNRW3FY3NXQCB4KPR3SODKFVANCNFSM4THGGZMA > > > >> . > >> > > > > — > You are receiving this because you authored the thread. > Reply to this email directly, view it on GitHub > < #29 (comment) >, > or unsubscribe > < https://github.com/notifications/unsubscribe-auth/AQ5AKFNZGKV2FV26Z6D5TMTSOZP2DANCNFSM4THGGZMA > > . > — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#29 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AA6BZDP25SSPVKVPN6BOBMTSO6KSLANCNFSM4THGGZMA> .