Welcome to the Kuzushiji Recognition Challenge!
\n\nYour task is to build a model that can read in pages from books written in ancient cursive Kuzushiji script, and localise and recognise characters and their modern japanese counterparts. With your help, CODH will be able to unlock the secrets of millions of books and make them accessible for all future Japanese generations, where they would otherwise be lost to time.
\n\nWe've spent many months preparing the data for this competition, so we hope you have a good time competing. Check out the visualisation kernel which lets you visualise the dataset as well as your submission files, and the Python implementation of our F-1 metric, which is slightly adapted from normal F-1 due to the nature of the competition labels.
\n\nGood luck, and happy Kaggling!
", "messages": [ { "id": 579530, "postDate": "2019-07-18T23:45:07.157Z", "content": "Welcome to the Kuzushiji Recognition Challenge!
\n\nYour task is to build a model that can read in pages from books written in ancient cursive Kuzushiji script, and localise and recognise characters and their modern japanese counterparts. With your help, CODH will be able to unlock the secrets of millions of books and make them accessible for all future Japanese generations, where they would otherwise be lost to time.
\n\nWe've spent many months preparing the data for this competition, so we hope you have a good time competing. Check out the visualisation kernel which lets you visualise the dataset as well as your submission files, and the Python implementation of our F-1 metric, which is slightly adapted from normal F-1 due to the nature of the competition labels.
\n\nGood luck, and happy Kaggling!
", "rawMarkdown": "Welcome to the Kuzushiji Recognition Challenge!\n\nYour task is to build a model that can read in pages from books written in ancient cursive Kuzushiji script, and localise and recognise characters and their modern japanese counterparts. With your help, CODH will be able to unlock the secrets of millions of books and make them accessible for all future Japanese generations, where they would otherwise be lost to time.\n\nWe've spent many months preparing the data for this competition, so we hope you have a good time competing. Check out the [visualisation kernel](https://www.kaggle.com/anokas/kuzushiji-visualisation) which lets you visualise the dataset as well as your submission files, and the [Python implementation of our F-1 metric](https://www.kaggle.com/anokas/kuzushiji-f-1-score), which is slightly adapted from normal F-1 due to the nature of the competition labels.\n\nGood luck, and happy Kaggling!", "votes": 58 }, { "id": 579542, "postDate": "2019-07-19T00:16:36.270Z", "content": "Thank you so much @anokas This competition is collaborative project between the Center for Open Data in the Humanities, the National Institute of Informatics and the National Institute of Japanese Literature in Japan. We hope the competition will help us find great algorithms to unlock the ancient texts which will potentially lead to new discovery in Japanese culture!\nThank you so much to all Kaggle staffs to make it happen!
", "rawMarkdown": "Thank you so much @anokas This competition is collaborative project between the Center for Open Data in the Humanities, the National Institute of Informatics and the National Institute of Japanese Literature in Japan. We hope the competition will help us find great algorithms to unlock the ancient texts which will potentially lead to new discovery in Japanese culture!\nThank you so much to all Kaggle staffs to make it happen! ", "votes": 21, "replies": [ { "id": 579546, "postDate": "2019-07-19T00:34:12.997Z", "content": "Thanks for the Competition Host. Nice Topic.
", "rawMarkdown": "Thanks for the Competition Host. Nice Topic.", "votes": 4 }, { "id": 586780, "postDate": "2019-07-29T17:12:30.743Z", "content": "Thanks for the Center for Open Data in the Humanities, the National Institute of Informatics and the National Institute of Japanese Literature in Japan for organizing ! Rediscovering hidden knowledge with data science is awesome !
", "rawMarkdown": "Thanks for the Center for Open Data in the Humanities, the National Institute of Informatics and the National Institute of Japanese Literature in Japan for organizing ! Rediscovering hidden knowledge with data science is awesome !", "votes": 1 } ] }, { "id": 579576, "postDate": "2019-07-19T01:54:56.497Z", "content": "Wow this looks so cool!\nA pro competitor organising a great comp for the community!
", "rawMarkdown": "Wow this looks so cool!\nA pro competitor organising a great comp for the community!", "votes": 6 }, { "id": 594452, "postDate": "2019-08-08T03:45:33.593Z", "content": "Thank you to CODH, NII, NIJL, @tkasasagi and @anokas for offering this competition. While in undergrad I spent a summer researching paleography during a period when the Castilian language was removed from Latin but not quite modern Spanish, and all of the graphical and other efforts were done by hand. I'm in this to learn and I'm fascinated by the possibilities that the final model could open for understanding the past.
", "rawMarkdown": "Thank you to CODH, NII, NIJL, @tkasasagi and @anokas for offering this competition. While in undergrad I spent a summer researching paleography during a period when the Castilian language was removed from Latin but not quite modern Spanish, and all of the graphical and other efforts were done by hand. I'm in this to learn and I'm fascinated by the possibilities that the final model could open for understanding the past.", "votes": 3 }, { "id": 579871, "postDate": "2019-07-19T10:35:51.027Z", "content": "Great kernal with the font visualizations for prediction results
", "rawMarkdown": "Great kernal with the font visualizations for prediction results", "votes": 3 }, { "id": 579969, "postDate": "2019-07-19T13:34:40.460Z", "content": "Thanks and congrats @anokas and your co-organizers for this hard and extensive work.
\n\nSuch huge and rich dataset is a tremendous contribution to the kaggle community.
", "rawMarkdown": "Thanks and congrats @anokas and your co-organizers for this hard and extensive work. \n\nSuch huge and rich dataset is a tremendous contribution to the kaggle community.", "votes": 4 }, { "id": 586749, "postDate": "2019-07-29T16:15:11.123Z", "content": "Thank you for hosting! As a Japanese, i can't miss this competition.
", "rawMarkdown": "Thank you for hosting! As a Japanese, i can't miss this competition.", "votes": 1 }, { "id": 583750, "postDate": "2019-07-24T23:23:27.310Z", "content": "Thank you for hosting this and preparing the data!\nThis competition is interesting theme for Japanese. It sounds a chance taking advantage of domain knowledge as Kanji or Hiragana. I will enjoy to try!
", "rawMarkdown": "Thank you for hosting this and preparing the data!\nThis competition is interesting theme for Japanese. It sounds a chance taking advantage of domain knowledge as Kanji or Hiragana. I will enjoy to try!", "votes": 1 }, { "id": 582031, "postDate": "2019-07-22T17:20:46.443Z", "content": "Wow.. this sounds interesting!
", "rawMarkdown": "Wow.. this sounds interesting!", "votes": 1 }, { "id": 581680, "postDate": "2019-07-22T08:46:33.163Z", "content": "Wow this looks so awesome!
", "rawMarkdown": "Wow this looks so awesome!", "votes": 1 }, { "id": 580362, "postDate": "2019-07-20T03:37:24.173Z", "content": "Very excited about the comp! I'm not an expert so I will just try to learn everything from you guys and see how far can we get to unlock the mystery of ancient Japan :)
", "rawMarkdown": "Very excited about the comp! I'm not an expert so I will just try to learn everything from you guys and see how far can we get to unlock the mystery of ancient Japan :)", "votes": 1 }, { "id": 580269, "postDate": "2019-07-19T23:13:52.843Z", "content": "Nice and challenging competition for font visualizations for prediction results
", "rawMarkdown": "Nice and challenging competition for font visualizations for prediction results", "votes": 1 }, { "id": 580217, "postDate": "2019-07-19T21:17:21.153Z", "content": "This is definitely a very valuable work.\ncongratulations for it
", "rawMarkdown": "\nThis is definitely a very valuable work.\ncongratulations for it", "votes": 1 }, { "id": 580140, "postDate": "2019-07-19T18:18:23.043Z", "content": "This is really cool!
", "rawMarkdown": "This is really cool!", "votes": 1 }, { "id": 580012, "postDate": "2019-07-19T14:41:58.617Z", "content": "This is interesting
", "rawMarkdown": "This is interesting", "votes": 1 }, { "id": 826303, "postDate": "2020-04-29T15:07:48.453Z", "content": "I found this competition 6 months later after it has ended. However I want to do late submission to validate my model performance. What confused me is that the note in submission page indicates the csv should have 4150 prediction rows while I can only find 1730 samples in test zip file as well as sample submission file. Any advices that can help me to make my submission correctly ?
", "rawMarkdown": "I found this competition 6 months later after it has ended. However I want to do late submission to validate my model performance. What confused me is that the note in submission page indicates the csv should have 4150 prediction rows while I can only find 1730 samples in test zip file as well as sample submission file. Any advices that can help me to make my submission correctly ?" }, { "id": 622624, "postDate": "2019-09-09T22:39:11.760Z", "content": "Thank you to host this competition!! I'm pleased to see the Japanese characters in data science competitions. And I truly appreciate the kuzushiji researchers to give all the annotated data for this classification task.
", "rawMarkdown": "Thank you to host this competition!! I'm pleased to see the Japanese characters in data science competitions. And I truly appreciate the kuzushiji researchers to give all the annotated data for this classification task. " }, { "id": 617764, "postDate": "2019-09-04T12:54:32.467Z", "content": "Hey, thanks for this competition. Am I allowed to create a dataset from cutouts of all character in the competition data?
", "rawMarkdown": "Hey, thanks for this competition. Am I allowed to create a dataset from cutouts of all character in the competition data?" }, { "id": 605878, "postDate": "2019-08-22T21:50:16.240Z", "content": "Thank you for such a challenge. Do we have entire metadata for identifying this language?
", "rawMarkdown": "Thank you for such a challenge. Do we have entire metadata for identifying this language?", "replies": [ { "id": 606232, "postDate": "2019-08-23T11:04:55.923Z", "content": "I'm not sure what you mean by metadata, could you rephrase?
", "rawMarkdown": "I'm not sure what you mean by metadata, could you rephrase?" } ] }, { "id": 590390, "postDate": "2019-08-02T06:37:39.757Z", "content": "This competition is so interesting! Thank you for your hosting! Ofcourse, I can't miss it.
", "rawMarkdown": "This competition is so interesting! Thank you for your hosting! Ofcourse, I can't miss it." }, { "id": 582206, "postDate": "2019-07-22T22:40:07.190Z", "content": "Cool topic, but really cheap payouts. For 3 months of hard work, if you are in the top 5, you can get a \"research grant\" worth roughly a month's pay at an entry level job(western USA)?
", "rawMarkdown": "Cool topic, but really cheap payouts. For 3 months of hard work, if you are in the top 5, you can get a \"research grant\" worth roughly a month's pay at an entry level job(western USA)?", "replies": [ { "id": 582211, "postDate": "2019-07-22T22:50:00.820Z", "content": "It's not my place to comment more generally, but I want to clarify that there is no limitation to use the money for research:
\n\nhttps://www.kaggle.com/c/kuzushiji-recognition/discussion/100561#580549
", "rawMarkdown": "It's not my place to comment more generally, but I want to clarify that there is no limitation to use the money for research: \n\nhttps://www.kaggle.com/c/kuzushiji-recognition/discussion/100561#580549", "votes": 3 } ] }, { "id": 656519, "postDate": "2019-10-24T10:44:23.853Z", "rawMarkdown": "", "isDeleted": true }, { "id": 590400, "postDate": "2019-08-02T06:48:20.597Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 579542, "author_name": "tkasasagi", "author_url": "", "post_date": "2019-07-19T00:16:36.270000", "content": "Thank you so much @anokas This competition is collaborative project between the Center for Open Data in the Humanities, the National Institute of Informatics and the National Institute of Japanese Literature in Japan. We hope the competition will help us find great algorithms to unlock the ancient texts which will potentially lead to new discovery in Japanese culture!\nThank you so much to all Kaggle staffs to make it happen!
", "votes": 21, "replies": [ { "id": 579546, "author_name": "Naruhiko Nakanishi", "author_url": "", "post_date": "2019-07-19T00:34:12.997000", "content": "Thanks for the Competition Host. Nice Topic.
", "votes": 4, "replies": [] }, { "id": 586780, "author_name": "", "author_url": "", "post_date": "2019-07-29T17:12:30.743000", "content": "Thanks for the Center for Open Data in the Humanities, the National Institute of Informatics and the National Institute of Japanese Literature in Japan for organizing ! Rediscovering hidden knowledge with data science is awesome !
", "votes": 1, "replies": [] } ] }, { "id": 579576, "author_name": "Aditya Soni", "author_url": "", "post_date": "2019-07-19T01:54:56.497000", "content": "Wow this looks so cool!\nA pro competitor organising a great comp for the community!
", "votes": 6, "replies": [] }, { "id": 594452, "author_name": "JeremyLoscheider", "author_url": "", "post_date": "2019-08-08T03:45:33.593000", "content": "Thank you to CODH, NII, NIJL, @tkasasagi and @anokas for offering this competition. While in undergrad I spent a summer researching paleography during a period when the Castilian language was removed from Latin but not quite modern Spanish, and all of the graphical and other efforts were done by hand. I'm in this to learn and I'm fascinated by the possibilities that the final model could open for understanding the past.
", "votes": 3, "replies": [] }, { "id": 579871, "author_name": "Ashik M", "author_url": "", "post_date": "2019-07-19T10:35:51.027000", "content": "Great kernal with the font visualizations for prediction results
", "votes": 3, "replies": [] }, { "id": 579969, "author_name": "Serigne ", "author_url": "", "post_date": "2019-07-19T13:34:40.460000", "content": "Thanks and congrats @anokas and your co-organizers for this hard and extensive work.
\n\nSuch huge and rich dataset is a tremendous contribution to the kaggle community.
", "votes": 4, "replies": [] }, { "id": 586749, "author_name": "Saira", "author_url": "", "post_date": "2019-07-29T16:15:11.123000", "content": "Thank you for hosting! As a Japanese, i can't miss this competition.
", "votes": 1, "replies": [] }, { "id": 583750, "author_name": "A.Umeda", "author_url": "", "post_date": "2019-07-24T23:23:27.310000", "content": "Thank you for hosting this and preparing the data!\nThis competition is interesting theme for Japanese. It sounds a chance taking advantage of domain knowledge as Kanji or Hiragana. I will enjoy to try!
", "votes": 1, "replies": [] }, { "id": 582031, "author_name": "Pramodini V Nayak", "author_url": "", "post_date": "2019-07-22T17:20:46.443000", "content": "Wow.. this sounds interesting!
", "votes": 1, "replies": [] }, { "id": 581680, "author_name": "Mani Bharathy C", "author_url": "", "post_date": "2019-07-22T08:46:33.163000", "content": "Wow this looks so awesome!
", "votes": 1, "replies": [] }, { "id": 580362, "author_name": "LeeMeng", "author_url": "", "post_date": "2019-07-20T03:37:24.173000", "content": "Very excited about the comp! I'm not an expert so I will just try to learn everything from you guys and see how far can we get to unlock the mystery of ancient Japan :)
", "votes": 1, "replies": [] }, { "id": 580269, "author_name": "Ishmeet singh", "author_url": "", "post_date": "2019-07-19T23:13:52.843000", "content": "Nice and challenging competition for font visualizations for prediction results
", "votes": 1, "replies": [] }, { "id": 580217, "author_name": "batu bayraktar", "author_url": "", "post_date": "2019-07-19T21:17:21.153000", "content": "This is definitely a very valuable work.\ncongratulations for it
", "votes": 1, "replies": [] }, { "id": 580140, "author_name": "Giovani Leone", "author_url": "", "post_date": "2019-07-19T18:18:23.043000", "content": "This is really cool!
", "votes": 1, "replies": [] }, { "id": 580012, "author_name": "Ahmed Melih Gokcel", "author_url": "", "post_date": "2019-07-19T14:41:58.617000", "content": "This is interesting
", "votes": 1, "replies": [] }, { "id": 826303, "author_name": "Keshawn Hsieh", "author_url": "", "post_date": "2020-04-29T15:07:48.453000", "content": "I found this competition 6 months later after it has ended. However I want to do late submission to validate my model performance. What confused me is that the note in submission page indicates the csv should have 4150 prediction rows while I can only find 1730 samples in test zip file as well as sample submission file. Any advices that can help me to make my submission correctly ?
", "votes": 0, "replies": [] }, { "id": 622624, "author_name": "Kenichi94", "author_url": "", "post_date": "2019-09-09T22:39:11.760000", "content": "Thank you to host this competition!! I'm pleased to see the Japanese characters in data science competitions. And I truly appreciate the kuzushiji researchers to give all the annotated data for this classification task.
", "votes": 0, "replies": [] }, { "id": 617764, "author_name": "Chris Wallenwein", "author_url": "", "post_date": "2019-09-04T12:54:32.467000", "content": "Hey, thanks for this competition. Am I allowed to create a dataset from cutouts of all character in the competition data?
", "votes": 0, "replies": [] }, { "id": 605878, "author_name": "Pegasus", "author_url": "", "post_date": "2019-08-22T21:50:16.240000", "content": "Thank you for such a challenge. Do we have entire metadata for identifying this language?
", "votes": 0, "replies": [ { "id": 606232, "author_name": "anokas", "author_url": "", "post_date": "2019-08-23T11:04:55.923000", "content": "I'm not sure what you mean by metadata, could you rephrase?
", "votes": 0, "replies": [] } ] }, { "id": 590390, "author_name": "Thomas Nguyen", "author_url": "", "post_date": "2019-08-02T06:37:39.757000", "content": "This competition is so interesting! Thank you for your hosting! Ofcourse, I can't miss it.
", "votes": 0, "replies": [] }, { "id": 582206, "author_name": "Ben Nye", "author_url": "", "post_date": "2019-07-22T22:40:07.190000", "content": "Cool topic, but really cheap payouts. For 3 months of hard work, if you are in the top 5, you can get a \"research grant\" worth roughly a month's pay at an entry level job(western USA)?
", "votes": 0, "replies": [ { "id": 582211, "author_name": "TheNuttyNetter", "author_url": "", "post_date": "2019-07-22T22:50:00.820000", "content": "It's not my place to comment more generally, but I want to clarify that there is no limitation to use the money for research:
\n\nhttps://www.kaggle.com/c/kuzushiji-recognition/discussion/100561#580549
", "votes": 3, "replies": [] } ] }, { "id": 656519, "author_name": "", "author_url": "", "post_date": "2019-10-24T10:44:23.853000", "content": "", "votes": 0, "replies": [] }, { "id": 590400, "author_name": "", "author_url": "", "post_date": "2019-08-02T06:48:20.597000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "579530": "Welcome to the Kuzushiji Recognition Challenge!\n\nYour task is to build a model that can read in pages from books written in ancient cursive Kuzushiji script, and localise and recognise characters and their modern japanese counterparts. With your help, CODH will be able to unlock the secrets of millions of books and make them accessible for all future Japanese generations, where they would otherwise be lost to time.\n\nWe've spent many months preparing the data for this competition, so we hope you have a good time competing. Check out the [visualisation kernel](https://www.kaggle.com/anokas/kuzushiji-visualisation) which lets you visualise the dataset as well as your submission files, and the [Python implementation of our F-1 metric](https://www.kaggle.com/anokas/kuzushiji-f-1-score), which is slightly adapted from normal F-1 due to the nature of the competition labels.\n\nGood luck, and happy Kaggling!", "579542": "Thank you so much @anokas This competition is collaborative project between the Center for Open Data in the Humanities, the National Institute of Informatics and the National Institute of Japanese Literature in Japan. We hope the competition will help us find great algorithms to unlock the ancient texts which will potentially lead to new discovery in Japanese culture!\nThank you so much to all Kaggle staffs to make it happen! ", "579576": "Wow this looks so cool!\nA pro competitor organising a great comp for the community!", "594452": "Thank you to CODH, NII, NIJL, @tkasasagi and @anokas for offering this competition. While in undergrad I spent a summer researching paleography during a period when the Castilian language was removed from Latin but not quite modern Spanish, and all of the graphical and other efforts were done by hand. I'm in this to learn and I'm fascinated by the possibilities that the final model could open for understanding the past.", "579871": "Great kernal with the font visualizations for prediction results", "579969": "Thanks and congrats @anokas and your co-organizers for this hard and extensive work. \n\nSuch huge and rich dataset is a tremendous contribution to the kaggle community.", "586749": "Thank you for hosting! As a Japanese, i can't miss this competition.", "583750": "Thank you for hosting this and preparing the data!\nThis competition is interesting theme for Japanese. It sounds a chance taking advantage of domain knowledge as Kanji or Hiragana. I will enjoy to try!", "582031": "Wow.. this sounds interesting!", "581680": "Wow this looks so awesome!", "580362": "Very excited about the comp! I'm not an expert so I will just try to learn everything from you guys and see how far can we get to unlock the mystery of ancient Japan :)", "580269": "Nice and challenging competition for font visualizations for prediction results", "580217": "\nThis is definitely a very valuable work.\ncongratulations for it", "580140": "This is really cool!", "580012": "This is interesting", "826303": "I found this competition 6 months later after it has ended. However I want to do late submission to validate my model performance. What confused me is that the note in submission page indicates the csv should have 4150 prediction rows while I can only find 1730 samples in test zip file as well as sample submission file. Any advices that can help me to make my submission correctly ?", "622624": "Thank you to host this competition!! I'm pleased to see the Japanese characters in data science competitions. And I truly appreciate the kuzushiji researchers to give all the annotated data for this classification task. ", "617764": "Hey, thanks for this competition. Am I allowed to create a dataset from cutouts of all character in the competition data?", "605878": "Thank you for such a challenge. Do we have entire metadata for identifying this language?", "590390": "This competition is so interesting! Thank you for your hosting! Ofcourse, I can't miss it.", "582206": "Cool topic, but really cheap payouts. For 3 months of hard work, if you are in the top 5, you can get a \"research grant\" worth roughly a month's pay at an entry level job(western USA)?", "656519": "", "590400": "" } }