![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F4910466%2F4477e1c70a30b0f96006ea57b3170462%2Flandmarks_7.gif?generation=1678879168458439&alt=media\")
Dear Kagglers,
\nWelcome to the Isolated Sign Language Recognition competition, where you can enhance PopSign's educational games for learning American Sign Language (ASL) - for everyone!
\nDid you know that Deaf children are at high risk of language deprivation syndrome? More than 90 percent of Deaf children are born to hearing parents who do not know sign language, which means they do not have early access to communication with their immediate family members. Studies indicate that language deprivation occurring within the first five years of Deaf children’s lives is a leading cause of lifelong reading, academic, employment, and social struggles.
\nYour participation will increase the ability for Deaf children to acquire language and connect with their family and community through effective communication. The ultimate goal is to develop a parent resource that is widely available and empowers parents to provide visual language to support their child’s growth towards kindergarten readiness and grade level standards.
\nThe PopSign app is an educational interactive bubble shooter mobile game, which inspires hearing parents or anyone to learn basic signs. This is an exciting opportunity to use an advanced recognition model that can detect visually complex movements of the hands used in ASL at a fast pace. Computer vision technology is ready for Sign Language, and you can contribute to taking ASL recognition to the next level.
\nAcquisition of language from birth is a universal human right for every child. When provided with access and opportunities, you build a bridge to enable communication between Deaf children and the world! Through enhancing the PopSign educational game, you will contribute to providing a powerful tool to make this a reality.
\nPlease enjoy, have fun, thank you, and the best of luck!
\nSam Sepah on behalf of the organizing team
", "messages": [ { "id": 2157221, "postDate": "2023-02-23T19:52:00.843Z", "content": "Dear Kagglers,
\nWelcome to the Isolated Sign Language Recognition competition, where you can enhance PopSign's educational games for learning American Sign Language (ASL) - for everyone!
\nDid you know that Deaf children are at high risk of language deprivation syndrome? More than 90 percent of Deaf children are born to hearing parents who do not know sign language, which means they do not have early access to communication with their immediate family members. Studies indicate that language deprivation occurring within the first five years of Deaf children’s lives is a leading cause of lifelong reading, academic, employment, and social struggles.
\nYour participation will increase the ability for Deaf children to acquire language and connect with their family and community through effective communication. The ultimate goal is to develop a parent resource that is widely available and empowers parents to provide visual language to support their child’s growth towards kindergarten readiness and grade level standards.
\nThe PopSign app is an educational interactive bubble shooter mobile game, which inspires hearing parents or anyone to learn basic signs. This is an exciting opportunity to use an advanced recognition model that can detect visually complex movements of the hands used in ASL at a fast pace. Computer vision technology is ready for Sign Language, and you can contribute to taking ASL recognition to the next level.
\nAcquisition of language from birth is a universal human right for every child. When provided with access and opportunities, you build a bridge to enable communication between Deaf children and the world! Through enhancing the PopSign educational game, you will contribute to providing a powerful tool to make this a reality.
\nPlease enjoy, have fun, thank you, and the best of luck!
\nSam Sepah on behalf of the organizing team
", "rawMarkdown": "Dear Kagglers,\n\nWelcome to the Isolated Sign Language Recognition competition, where you can enhance PopSign's educational games for learning American Sign Language (ASL) - for everyone! \n\nDid you know that Deaf children are at high risk of language deprivation syndrome? More than 90 percent of Deaf children are born to hearing parents who do not know sign language, which means they do not have early access to communication with their immediate family members. Studies indicate that language deprivation occurring within the first five years of Deaf children’s lives is a leading cause of lifelong reading, academic, employment, and social struggles.\n\nYour participation will increase the ability for Deaf children to acquire language and connect with their family and community through effective communication. The ultimate goal is to develop a parent resource that is widely available and empowers parents to provide visual language to support their child’s growth towards kindergarten readiness and grade level standards. \n\nThe PopSign app is an educational interactive bubble shooter mobile game, which inspires hearing parents or anyone to learn basic signs. This is an exciting opportunity to use an advanced recognition model that can detect visually complex movements of the hands used in ASL at a fast pace. Computer vision technology is ready for Sign Language, and you can contribute to taking ASL recognition to the next level. \n\nAcquisition of language from birth is a universal human right for every child. When provided with access and opportunities, you build a bridge to enable communication between Deaf children and the world! Through enhancing the PopSign educational game, you will contribute to providing a powerful tool to make this a reality. \n\n\nPlease enjoy, have fun, thank you, and the best of luck!\n\nSam Sepah on behalf of the organizing team\n", "votes": 28 }, { "id": 2169443, "postDate": "2023-03-05T05:46:08.100Z", "content": "Hi @samsepah, thanks for hosting an interesting and unique competition!
\nLet me ask about the external dataset. There are a lot of sign language pose dataset like WLASL, is there any license requirement?
\nThe rule jsut says the external dataset should be publicly available, but some competitions require they have to be commercially usable. I wonder how about in this competition.
Thanks in advance!
", "rawMarkdown": "Hi @samsepah, thanks for hosting an interesting and unique competition! \nLet me ask about the external dataset. There are a lot of sign language pose dataset like [WLASL](https://github.com/dxli94/WLASL), is there any license requirement? \nThe rule jsut says the external dataset should be publicly available, but some competitions require they have to be commercially usable. I wonder how about in this competition.\n\nThanks in advance!\n", "votes": 7, "replies": [ { "id": 2175500, "postDate": "2023-03-09T22:14:16.350Z", "content": "Thanks Camaro for your question! External Data needs to allow you to license the winning solution, i.e. being used in commercial purpose.
\nHere below please see the relevant rules for details:
\n\n", "rawMarkdown": "Thanks Camaro for your question! External Data needs to allow you to license the winning solution, i.e. being used in commercial purpose. \n\nHere below please see the relevant rules for details:\n>1. WINNER LICENSE.\nUnder Section 11 (Winners Obligations) of the General Rules below, you hereby grant and will grant Competition Sponsor the following license(s) with respect to your Submission if you are a Competition winner:\nOpen Source: You hereby license and will license your winning Submission and the source code used to generate the Submission under an Apache 2.0 License that in no event limits commercial use of such code or model containing or depending on such code. To the extent your Submission makes use of generally commercially available software not owned by you that you used to generate your submission, but that can be procured by the Competition Sponsor without undue expense, you do not grant the license in the preceding sentence to that software.\n...\n>C. 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>11. WINNERS OBLIGATIONS.\nAs a condition to being awarded a Prize, a Prize winner must fulfill the following obligations:\n(a) deliver to the Competition Sponsor the final model's software code as used to generate the winning Submission and associated documentation. The delivered software code should follow these documentation guidelines, must be capable of generating the winning Submission, and contain a description of resources required to build and/or run the executable code successfully. To the extent that the final model’s software code includes generally commercially available software that is not owned by you, but that can be procured by the Competition Sponsor without undue expense, then instead of delivering the code for that software to the Competition Sponsor, you must identify that software, method for procuring it, and any parameters or other information necessary to replicate the winning Submission;\n(b) grant to the Competition Sponsor the license to the winning Submission stated in the Competition Specific Rules above, and represent that you have the unrestricted right to grant that license", "votes": 3, "replies": [ { "id": 2198036, "postDate": "2023-03-26T16:28:14.443Z", "content": "\n
\n- WINNER LICENSE.
\n
\n Under Section 11 (Winners Obligations) of the General Rules below, you hereby grant and will grant Competition Sponsor the following license(s) with respect to your Submission if you are a Competition winner:
\n Open Source: You hereby license and will license your winning Submission and the source code used to generate the Submission under an Apache 2.0 License that in no event limits commercial use of such code or model containing or depending on such code. To the extent your Submission makes use of generally commercially available software not owned by you that you used to generate your submission, but that can be procured by the Competition Sponsor without undue expense, you do not grant the license in the preceding sentence to that software.
\n …
\n C. 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 …- WINNERS OBLIGATIONS.
\n
\n As a condition to being awarded a Prize, a Prize winner must fulfill the following obligations:
\n (a) deliver to the Competition Sponsor the final model's software code as used to generate the winning Submission and associated documentation. The delivered software code should follow these documentation guidelines, must be capable of generating the winning Submission, and contain a description of resources required to build and/or run the executable code successfully. To the extent that the final model’s software code includes generally commercially available software that is not owned by you, but that can be procured by the Competition Sponsor without undue expense, then instead of delivering the code for that software to the Competition Sponsor, you must identify that software, method for procuring it, and any parameters or other information necessary to replicate the winning Submission;
\n (b) grant to the Competition Sponsor the license to the winning Submission stated in the Competition Specific Rules above, and represent that you have the unrestricted right to grant that license
Hi @ashleychow, I found a dataset that doesn't mention anything (doesnt permit or refrain directly) regarding commercial usage. It is a \"Computational Use of Data Agreement v0.1\", and it says that \"You agree that you will use the Data solely for Computational Use.\" I'm not sure if it is permitted for the competition. Can you please let me know, thanks!
", "rawMarkdown": "Hi @ashleychow, I found a dataset that doesn't mention anything (doesnt permit or refrain directly) regarding commercial usage. It is a \"Computational Use of Data Agreement v0.1\", and it says that \"You agree that you will use the Data solely for Computational Use.\" I'm not sure if it is permitted for the competition. Can you please let me know, thanks!" }, { "id": 2198110, "postDate": "2023-03-26T17:18:33.377Z", "content": "WLASL, MSASL, ASLLVD, etc. are all published under this C-UDA license.
\nHowever, commercial usage is not allowed under such a license, as stated on the WLASL homepage. [1]
Disclaimer\nAll the WLASL data is intended for academic and computational use only. No commercial usage is allowed. We highly respect copyright and privacy. If you find WLASL violates your rights, please contact us.\nTherefore I assume it is not allowed to use any of these datasets as external data or to use pre-trained model checkpoints that were trained on one of these datasets.
\nFeel free to correct me if I am wrong..
\n[1] https://dxli94.github.io/WLASL/
", "rawMarkdown": "WLASL, MSASL, ASLLVD, etc. are all published under this C-UDA license.\nHowever, commercial usage is not allowed under such a license, as stated on the WLASL homepage. [1]\n\n```\nDisclaimer\nAll the WLASL data is intended for academic and computational use only. No commercial usage is allowed. We highly respect copyright and privacy. If you find WLASL violates your rights, please contact us.\n```\n\nTherefore I assume it is not allowed to use any of these datasets as external data or to use pre-trained model checkpoints that were trained on one of these datasets. \n\nFeel free to correct me if I am wrong..\n\n[1] https://dxli94.github.io/WLASL/" }, { "id": 2201501, "postDate": "2023-03-29T11:08:35.550Z", "content": "Thanks @ashwinrameshasdsa! Unfortunately I am not a legal expert and so unable to provide any legal guidance on the dataset license interpretation. Thanks for your kind understanding!
", "rawMarkdown": "Thanks @ashwinrameshasdsa! Unfortunately I am not a legal expert and so unable to provide any legal guidance on the dataset license interpretation. Thanks for your kind understanding! " } ] } ] }, { "id": 2177000, "postDate": "2023-03-11T05:37:26.890Z", "content": "Hello!, thank you for making this competition and let us taking part in it.
\nI have a question, I've working in a similar problem for a while but i'd used 3D Render, and some tools from robotics like cinematic, dynamics, control. etc. I know that this way is far harder than using AI but I wanted to take a big challenge, I also need to take data (not in the same way as you. the data I take is also points in space but it isn't the entire information, just a tiny part of it). But I think I could use this data/information to develop my project and make sure that is working at it should.
\nCould I use competition data and information to integrate it into my project (I want it to be open source).
", "rawMarkdown": "Hello!, thank you for making this competition and let us taking part in it.\n\nI have a question, I've working in a similar problem for a while but i'd used 3D Render, and some tools from robotics like cinematic, dynamics, control. etc. I know that this way is far harder than using AI but I wanted to take a big challenge, I also need to take data (not in the same way as you. the data I take is also points in space but it isn't the entire information, just a tiny part of it). But I think I could use this data/information to develop my project and make sure that is working at it should.\n\nCould I use competition data and information to integrate it into my project (I want it to be open source).", "votes": 3 }, { "id": 2196867, "postDate": "2023-03-25T17:57:29.793Z", "content": "@samsepah Hello! Can we use our own videos for training?
\nIf I have my own small ASL dataset, I can make it public (and commercially available). Would it be permitted to use it for training in this case?
Hi @ilya9711nov, thanks for your question! Yes, for any external data, you may use it as long as it is public and commercially available. See relevant rules below. Thanks!
\n\n", "rawMarkdown": "Hi @ilya9711nov, thanks for your question! Yes, for any external data, you may use it as long as it is public and commercially available. See relevant rules below. Thanks!\n\n>C. 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).", "votes": 1 } ] }, { "id": 2182952, "postDate": "2023-03-15T11:57:45.477Z", "content": "C. 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
I noticed that the submission inference is quite slow compared to my local machine.
\nCan you disclose how the submission is processed? I'm guessing it uses CPUs for the inference, but since this submission process is new, it would be great to know about the specifics (especially the hardware)
", "rawMarkdown": "I noticed that the submission inference is quite slow compared to my local machine.\n\nCan you disclose how the submission is processed? I'm guessing it uses CPUs for the inference, but since this submission process is new, it would be great to know about the specifics (especially the hardware)", "votes": 1, "replies": [ { "id": 2186172, "postDate": "2023-03-17T16:25:33.257Z", "content": "you can measure the time on the public notebook.
\nThe timming are close for my LB submission.
use !lscpu for cpu information
", "rawMarkdown": "you can measure the time on the public notebook.\nThe timming are close for my LB submission.\n\nuse !lscpu for cpu information", "replies": [ { "id": 2186404, "postDate": "2023-03-17T18:57:56.940Z", "content": "that's too bad.
\nTflite is optimized for arm processors and not x86, so we cannot get the best out with
\nconverter.optimizations = [tf.lite.Optimize.DEFAULT]
\nYou actually see much slower operations with quantized models with x86 since it does not support INT8 operations
\nhttps://github.com/tensorflow/tensorflow/issues/21698
Nice job man.!
", "rawMarkdown": "Nice job man.!", "votes": -1 }, { "id": 2219233, "postDate": "2023-04-12T12:11:21.877Z", "content": "1st Kaggle！
", "rawMarkdown": "1st Kaggle！" }, { "id": 2211484, "postDate": "2023-04-06T04:20:05.040Z", "content": "I wish to join and contribute to a team.
", "rawMarkdown": "I wish to join and contribute to a team." }, { "id": 2206517, "postDate": "2023-04-02T16:02:49.527Z", "content": "Really excited about this Kaggle competition. I truly believe that it can positively impact the lives of many people around the world! Let's go!!
", "rawMarkdown": "Really excited about this Kaggle competition. I truly believe that it can positively impact the lives of many people around the world! Let's go!!" }, { "id": 2190132, "postDate": "2023-03-21T04:06:54.997Z", "content": "@samsepah Can we use videos on PopSign iOS app repository for training?
\nhttps://github.com/Accessible-Technology-in-Sign/PopSign
Hi @tatamikenn, thanks for your question! Yes, same as above question - you may use any external data as long as it is public and commercially available. Thanks so much!
", "rawMarkdown": "Hi @tatamikenn, thanks for your question! Yes, same as above question - you may use any external data as long as it is public and commercially available. Thanks so much!" } ] }, { "id": 2189282, "postDate": "2023-03-20T11:23:26.403Z", "content": "Hello!
\nThank you for making that competition! That's a great opportunity to dive into an adventurous research
", "rawMarkdown": "Hello! \n\nThank you for making that competition! That's a great opportunity to dive into an adventurous research" }, { "id": 2182898, "postDate": "2023-03-15T11:22:30.647Z", "content": "@samsepah
\nAnother question. In the landmark data, sometimes signers move their legs (sometimes with a large move).
\nWhy this occurs? In my understanding, sign language doesn't need to move legs.
\nOr, is this just MediaPipe's labeling error?
@samsepah I have a question.
\nWhat is the sampling rate of the landmark data?
\nI think this is important when combining with other external data.
hello !
\nThank you for hosting this competition and we are more than happy to take part!
Another question. Are train/test datasets split by participant_id?
\nIn other words, do train/test dataset have same participant_id? I guess not, but just want make sure.
This question was answered by Kaggle staff in another thread. [1]
\n\n\nThere are new participant IDs in the test set. However, your submission notebook does not have access to the participant IDs for any given video.
\n
[1] https://www.kaggle.com/competitions/asl-signs/discussion/390510#2171517
", "rawMarkdown": "This question was answered by Kaggle staff in another thread. [1]\n\n>There are new participant IDs in the test set. However, your submission notebook does not have access to the participant IDs for any given video.\n\n[1] https://www.kaggle.com/competitions/asl-signs/discussion/390510#2171517", "votes": 1 } ] }, { "id": 2174545, "postDate": "2023-03-09T07:54:03.787Z", "content": "Greetings :)
\nI had a quick question -
\nI read that the model should occupy at max 40mbs. Does that mean, all the dependencies should be under that? Alternatively can we have certain dependencies significantly exceeding 40MBswhich can come from pre processing the data for instance? The deep learning model will indeed be below 40Mbs. Thanks for taking time!
The constraint only applies to the model checkpoint file.
", "rawMarkdown": "The constraint only applies to the model checkpoint file.", "replies": [ { "id": 2175198, "postDate": "2023-03-09T18:04:33.847Z", "content": "Perfect! Thanks for your response :)
", "rawMarkdown": "Perfect! Thanks for your response :) " } ] } ] }, { "id": 2242558, "postDate": "2023-05-02T11:10:09.763Z", "rawMarkdown": "", "votes": 1, "isDeleted": true } ], "comments": [ { "id": 2169443, "author_name": "Camaro", "author_url": "", "post_date": "2023-03-05T05:46:08.100000", "content": "Hi @samsepah, thanks for hosting an interesting and unique competition!
\nLet me ask about the external dataset. There are a lot of sign language pose dataset like WLASL, is there any license requirement?
\nThe rule jsut says the external dataset should be publicly available, but some competitions require they have to be commercially usable. I wonder how about in this competition.
Thanks in advance!
", "votes": 7, "replies": [ { "id": 2175500, "author_name": "Ashley Chow", "author_url": "", "post_date": "2023-03-09T22:14:16.350000", "content": "Thanks Camaro for your question! External Data needs to allow you to license the winning solution, i.e. being used in commercial purpose.
\nHere below please see the relevant rules for details:
\n\n", "votes": 3, "replies": [ { "id": 2198036, "author_name": "Ashwin Ramesh", "author_url": "", "post_date": "2023-03-26T16:28:14.443000", "content": "\n
\n- WINNER LICENSE.
\n
\n Under Section 11 (Winners Obligations) of the General Rules below, you hereby grant and will grant Competition Sponsor the following license(s) with respect to your Submission if you are a Competition winner:
\n Open Source: You hereby license and will license your winning Submission and the source code used to generate the Submission under an Apache 2.0 License that in no event limits commercial use of such code or model containing or depending on such code. To the extent your Submission makes use of generally commercially available software not owned by you that you used to generate your submission, but that can be procured by the Competition Sponsor without undue expense, you do not grant the license in the preceding sentence to that software.
\n …
\n C. 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 …- WINNERS OBLIGATIONS.
\n
\n As a condition to being awarded a Prize, a Prize winner must fulfill the following obligations:
\n (a) deliver to the Competition Sponsor the final model's software code as used to generate the winning Submission and associated documentation. The delivered software code should follow these documentation guidelines, must be capable of generating the winning Submission, and contain a description of resources required to build and/or run the executable code successfully. To the extent that the final model’s software code includes generally commercially available software that is not owned by you, but that can be procured by the Competition Sponsor without undue expense, then instead of delivering the code for that software to the Competition Sponsor, you must identify that software, method for procuring it, and any parameters or other information necessary to replicate the winning Submission;
\n (b) grant to the Competition Sponsor the license to the winning Submission stated in the Competition Specific Rules above, and represent that you have the unrestricted right to grant that license
Hi @ashleychow, I found a dataset that doesn't mention anything (doesnt permit or refrain directly) regarding commercial usage. It is a \"Computational Use of Data Agreement v0.1\", and it says that \"You agree that you will use the Data solely for Computational Use.\" I'm not sure if it is permitted for the competition. Can you please let me know, thanks!
", "votes": 0, "replies": [] }, { "id": 2198110, "author_name": "Gregor Lied", "author_url": "", "post_date": "2023-03-26T17:18:33.377000", "content": "WLASL, MSASL, ASLLVD, etc. are all published under this C-UDA license.
\nHowever, commercial usage is not allowed under such a license, as stated on the WLASL homepage. [1]
Disclaimer\nAll the WLASL data is intended for academic and computational use only. No commercial usage is allowed. We highly respect copyright and privacy. If you find WLASL violates your rights, please contact us.\nTherefore I assume it is not allowed to use any of these datasets as external data or to use pre-trained model checkpoints that were trained on one of these datasets.
\nFeel free to correct me if I am wrong..
\n[1] https://dxli94.github.io/WLASL/
", "votes": 0, "replies": [] }, { "id": 2201501, "author_name": "Ashley Chow", "author_url": "", "post_date": "2023-03-29T11:08:35.550000", "content": "Thanks @ashwinrameshasdsa! Unfortunately I am not a legal expert and so unable to provide any legal guidance on the dataset license interpretation. Thanks for your kind understanding!
", "votes": 0, "replies": [] } ] } ] }, { "id": 2177000, "author_name": "Hanamori", "author_url": "", "post_date": "2023-03-11T05:37:26.890000", "content": "Hello!, thank you for making this competition and let us taking part in it.
\nI have a question, I've working in a similar problem for a while but i'd used 3D Render, and some tools from robotics like cinematic, dynamics, control. etc. I know that this way is far harder than using AI but I wanted to take a big challenge, I also need to take data (not in the same way as you. the data I take is also points in space but it isn't the entire information, just a tiny part of it). But I think I could use this data/information to develop my project and make sure that is working at it should.
\nCould I use competition data and information to integrate it into my project (I want it to be open source).
", "votes": 3, "replies": [] }, { "id": 2196867, "author_name": "Ilya Novoselskiy", "author_url": "", "post_date": "2023-03-25T17:57:29.793000", "content": "@samsepah Hello! Can we use our own videos for training?
\nIf I have my own small ASL dataset, I can make it public (and commercially available). Would it be permitted to use it for training in this case?
Hi @ilya9711nov, thanks for your question! Yes, for any external data, you may use it as long as it is public and commercially available. See relevant rules below. Thanks!
\n\n", "votes": 1, "replies": [] } ] }, { "id": 2182952, "author_name": "arutema47", "author_url": "", "post_date": "2023-03-15T11:57:45.477000", "content": "C. 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
I noticed that the submission inference is quite slow compared to my local machine.
\nCan you disclose how the submission is processed? I'm guessing it uses CPUs for the inference, but since this submission process is new, it would be great to know about the specifics (especially the hardware)
", "votes": 1, "replies": [ { "id": 2186172, "author_name": "hengck23", "author_url": "", "post_date": "2023-03-17T16:25:33.257000", "content": "you can measure the time on the public notebook.
\nThe timming are close for my LB submission.
use !lscpu for cpu information
", "votes": 0, "replies": [ { "id": 2186404, "author_name": "arutema47", "author_url": "", "post_date": "2023-03-17T18:57:56.940000", "content": "that's too bad.
\nTflite is optimized for arm processors and not x86, so we cannot get the best out with
\nconverter.optimizations = [tf.lite.Optimize.DEFAULT]
\nYou actually see much slower operations with quantized models with x86 since it does not support INT8 operations
\nhttps://github.com/tensorflow/tensorflow/issues/21698
Nice job man.!
", "votes": -1, "replies": [] }, { "id": 2219233, "author_name": "xiayu", "author_url": "", "post_date": "2023-04-12T12:11:21.877000", "content": "1st Kaggle！
", "votes": 0, "replies": [] }, { "id": 2211484, "author_name": "Yeakub Sadlil", "author_url": "", "post_date": "2023-04-06T04:20:05.040000", "content": "I wish to join and contribute to a team.
", "votes": 0, "replies": [] }, { "id": 2206517, "author_name": "George Theodorou", "author_url": "", "post_date": "2023-04-02T16:02:49.527000", "content": "Really excited about this Kaggle competition. I truly believe that it can positively impact the lives of many people around the world! Let's go!!
", "votes": 0, "replies": [] }, { "id": 2190132, "author_name": "Bilzard", "author_url": "", "post_date": "2023-03-21T04:06:54.997000", "content": "@samsepah Can we use videos on PopSign iOS app repository for training?
\nhttps://github.com/Accessible-Technology-in-Sign/PopSign
Hi @tatamikenn, thanks for your question! Yes, same as above question - you may use any external data as long as it is public and commercially available. Thanks so much!
", "votes": 0, "replies": [] } ] }, { "id": 2189282, "author_name": "Tokhchukov Danil", "author_url": "", "post_date": "2023-03-20T11:23:26.403000", "content": "Hello!
\nThank you for making that competition! That's a great opportunity to dive into an adventurous research
", "votes": 0, "replies": [] }, { "id": 2182898, "author_name": "Bilzard", "author_url": "", "post_date": "2023-03-15T11:22:30.647000", "content": "@samsepah
\nAnother question. In the landmark data, sometimes signers move their legs (sometimes with a large move).
\nWhy this occurs? In my understanding, sign language doesn't need to move legs.
\nOr, is this just MediaPipe's labeling error?
@samsepah I have a question.
\nWhat is the sampling rate of the landmark data?
\nI think this is important when combining with other external data.
hello !
\nThank you for hosting this competition and we are more than happy to take part!
Another question. Are train/test datasets split by participant_id?
\nIn other words, do train/test dataset have same participant_id? I guess not, but just want make sure.
This question was answered by Kaggle staff in another thread. [1]
\n\n\nThere are new participant IDs in the test set. However, your submission notebook does not have access to the participant IDs for any given video.
\n
[1] https://www.kaggle.com/competitions/asl-signs/discussion/390510#2171517
", "votes": 1, "replies": [] } ] }, { "id": 2174545, "author_name": "siddhant bahuguna", "author_url": "", "post_date": "2023-03-09T07:54:03.787000", "content": "Greetings :)
\nI had a quick question -
\nI read that the model should occupy at max 40mbs. Does that mean, all the dependencies should be under that? Alternatively can we have certain dependencies significantly exceeding 40MBswhich can come from pre processing the data for instance? The deep learning model will indeed be below 40Mbs. Thanks for taking time!
The constraint only applies to the model checkpoint file.
", "votes": 0, "replies": [ { "id": 2175198, "author_name": "siddhant bahuguna", "author_url": "", "post_date": "2023-03-09T18:04:33.847000", "content": "Perfect! Thanks for your response :)
", "votes": 0, "replies": [] } ] } ] }, { "id": 2242558, "author_name": "", "author_url": "", "post_date": "2023-05-02T11:10:09.763000", "content": "", "votes": 1, "replies": [] } ], "raw_markdown_by_id": { "2157221": "Dear Kagglers,\n\nWelcome to the Isolated Sign Language Recognition competition, where you can enhance PopSign's educational games for learning American Sign Language (ASL) - for everyone! \n\nDid you know that Deaf children are at high risk of language deprivation syndrome? More than 90 percent of Deaf children are born to hearing parents who do not know sign language, which means they do not have early access to communication with their immediate family members. Studies indicate that language deprivation occurring within the first five years of Deaf children’s lives is a leading cause of lifelong reading, academic, employment, and social struggles.\n\nYour participation will increase the ability for Deaf children to acquire language and connect with their family and community through effective communication. The ultimate goal is to develop a parent resource that is widely available and empowers parents to provide visual language to support their child’s growth towards kindergarten readiness and grade level standards. \n\nThe PopSign app is an educational interactive bubble shooter mobile game, which inspires hearing parents or anyone to learn basic signs. This is an exciting opportunity to use an advanced recognition model that can detect visually complex movements of the hands used in ASL at a fast pace. Computer vision technology is ready for Sign Language, and you can contribute to taking ASL recognition to the next level. \n\nAcquisition of language from birth is a universal human right for every child. When provided with access and opportunities, you build a bridge to enable communication between Deaf children and the world! Through enhancing the PopSign educational game, you will contribute to providing a powerful tool to make this a reality. \n\n\nPlease enjoy, have fun, thank you, and the best of luck!\n\nSam Sepah on behalf of the organizing team\n", "2169443": "Hi @samsepah, thanks for hosting an interesting and unique competition! \nLet me ask about the external dataset. There are a lot of sign language pose dataset like [WLASL](https://github.com/dxli94/WLASL), is there any license requirement? \nThe rule jsut says the external dataset should be publicly available, but some competitions require they have to be commercially usable. I wonder how about in this competition.\n\nThanks in advance!\n", "2177000": "Hello!, thank you for making this competition and let us taking part in it.\n\nI have a question, I've working in a similar problem for a while but i'd used 3D Render, and some tools from robotics like cinematic, dynamics, control. etc. I know that this way is far harder than using AI but I wanted to take a big challenge, I also need to take data (not in the same way as you. the data I take is also points in space but it isn't the entire information, just a tiny part of it). But I think I could use this data/information to develop my project and make sure that is working at it should.\n\nCould I use competition data and information to integrate it into my project (I want it to be open source).", "2196867": "@samsepah Hello! Can we use our own videos for training? \nIf I have my own small ASL dataset, I can make it public (and commercially available). Would it be permitted to use it for training in this case?", "2182952": "I noticed that the submission inference is quite slow compared to my local machine.\n\nCan you disclose how the submission is processed? I'm guessing it uses CPUs for the inference, but since this submission process is new, it would be great to know about the specifics (especially the hardware)", "2204342": "Nice job man.!", "2219233": "1st Kaggle！", "2211484": "I wish to join and contribute to a team.", "2206517": "Really excited about this Kaggle competition. I truly believe that it can positively impact the lives of many people around the world! Let's go!!", "2190132": "@samsepah Can we use videos on PopSign iOS app repository for training?\nhttps://github.com/Accessible-Technology-in-Sign/PopSign", "2189282": "Hello! \n\nThank you for making that competition! That's a great opportunity to dive into an adventurous research", "2182898": "@samsepah \nAnother question. In the landmark data, sometimes signers move their legs (sometimes with a large move).\nWhy this occurs? In my understanding, sign language doesn't need to move legs.\nOr, is this just MediaPipe's labeling error?\n\n", "2182876": "@samsepah I have a question.\n\nWhat is the sampling rate of the landmark data?\nI think this is important when combining with other external data.", "2178242": "hello !\nThank you for hosting this competition and we are more than happy to take part!", "2177254": "Another question. Are train/test datasets split by participant_id?\nIn other words, do train/test dataset have same participant_id? I guess not, but just want make sure.", "2174545": "Greetings :)\nI had a quick question -\nI read that the model should occupy at max 40mbs. Does that mean, all the dependencies should be under that? Alternatively can we have certain dependencies significantly exceeding 40MBswhich can come from pre processing the data for instance? The deep learning model will indeed be below 40Mbs. Thanks for taking time!", "2242558": "" } }