{"id":33125744,"url":"https://github.com/Jean-Baptiste-Camps/FROC-MSS","last_synced_at":"2025-11-20T00:02:07.828Z","repository":{"id":146917450,"uuid":"162571642","full_name":"Jean-Baptiste-Camps/FROC-MSS","owner":"Jean-Baptiste-Camps","description":"FROC-MSS: Old French and Old Occitan Medieval Manuscripts HTR Data and Models","archived":false,"fork":false,"pushed_at":"2019-01-08T14:52:11.000Z","size":28105,"stargazers_count":0,"open_issues_count":0,"forks_count":1,"subscribers_count":3,"default_branch":"master","last_synced_at":"2025-03-05T16:53:35.877Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":null,"language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/Jean-Baptiste-Camps.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null}},"created_at":"2018-12-20T11:40:05.000Z","updated_at":"2019-01-08T14:52:13.000Z","dependencies_parsed_at":"2024-01-07T21:24:23.079Z","dependency_job_id":null,"html_url":"https://github.com/Jean-Baptiste-Camps/FROC-MSS","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/Jean-Baptiste-Camps/FROC-MSS","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Jean-Baptiste-Camps%2FFROC-MSS","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Jean-Baptiste-Camps%2FFROC-MSS/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Jean-Baptiste-Camps%2FFROC-MSS/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Jean-Baptiste-Camps%2FFROC-MSS/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/Jean-Baptiste-Camps","download_url":"https://codeload.github.com/Jean-Baptiste-Camps/FROC-MSS/tar.gz/refs/heads/master","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Jean-Baptiste-Camps%2FFROC-MSS/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":285348528,"owners_count":27156475,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","status":"online","status_checked_at":"2025-11-19T02:00:05.673Z","response_time":65,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2025-11-15T06:00:52.704Z","updated_at":"2025-11-20T00:02:07.823Z","avatar_url":"https://github.com/Jean-Baptiste-Camps.png","language":"Python","funding_links":[],"categories":["Datasets"],"sub_categories":["Ground Truth"],"readme":"# FROC-MSS: Old French and Old Occitan Medieval Manuscripts HTR Data and Models\n\nThis repository contains:\n\n- training and evaluation data from allographetic transcriptions of various\nOld French and Old Occitan manuscripts, in various states of correctness, in Kraken\ntraining format;\n- HTR models trained and tested using this data.\n\nIf you plan of using this data or the provided model for a publication, please\ncite it, as:\n\nJean-Baptiste Camps (éd.), _FROC-MSS: Old French and Old Occitan Medieval Manuscripts HTR Data and Models_, Paris: École nationale des chartes (PSL), 2018, [https://github.com/Jean-Baptiste-Camps/FROC-MSS](https://github.com/Jean-Baptiste-Camps/FROC-MSS).\n\n## Data format\n\nThe data is as following:\n- each line image in a `.png` file;\n- each transcription in a `.gt.txt` file.\n\nUnicode NFD normalisation has been applied on the ground-truth text.\n\n## Models\n\n### Summary and C.E.R.\n\nThe root folder contains a vanilla Kraken model (`model_froc.mlmodel`), \ntrained with default settings and without any additional data (e.g. no artificial noised data).\n\nData was randomly divided in 80\\% for training (`train.txt`), 10\\% for in-training \nvalidation (`val.txt`) and 10\\% for final testing of the model (`test.txt`).\n\nIt achieved a C.E.R. of:\n\n- ** 8.11 % ** on validation data (7.03% ignoring spaces);\n- ** 7.83 % ** on test data (6.92% ignoring spaces).\n\n### Errors and most frequent confusions on test data\n\nThere were 13540 characters and 1061 errors on test data.\n\nGlobally, the error are as follow: \n\n- 536 characters from the ground truth were not predicted by the model;\n- 132 characters absent from the ground truth were wrongly predicted;\n- 393 character substitutions.\n\n\nThe most frequent confusions concerned spacing.\n\nThe 20 most frequent confusions are:\n\n    Errors\tGround Truth-Prediction\n    70\t{ SPACE } - {  }\n    54\t{  } - { SPACE }\n    48\t{ ı } - {  }\n    43\t{ n } - {  }\n    43\t{ COMBINING ACUTE ACCENT } - {  }\n    27\t{ e } - {  }\n    24\t{ l } - {  }\n    24\t{ u } - {  }\n    21\t{ . } - {  }\n    20\t{ u } - { n }\n    18\t{ ſ } - {  }\n    18\t{ a } - {  }\n    17\t{ r } - {  }\n    14\t{ t } - {  }\n    13\t{ COMBINING TILDE } - {  }\n    13\t{  } - { ı }\n    12\t{ o } - { e }\n    12\t{ o } - {  }\n    12\t{ ı } - { m }\n    11\t{ e } - { c }\n\n\n## List of manuscripts\n\nThe data comes from partial allographetic transcription of the following mss:\n\n- Clermont-Ferrand, archives départementales, 1F2 (XIII 1/3, anglo-norman _praegothica_ script; _Chanson d'Aspremont_); 52 lines.\n- Paris, Bibliothèque nationale de France, fr. 854 (XIII 4/4, Venice or Venetian area; gothic _textualis_;  occitan chansonnier _I_); 1112 lines.\n- Cologny-Genève, fondation Martin-Bodmer, cod. Bodm. 168 (XIII 3/3, anglo-norman gothic _textualis_; _Chanson d'Otinel_); 1908 lines.\n- Oxford, Bodleian Library, Digby 23 (XII 1/2, anglo-norman _praegothica_; _Chanson de Roland_); 564 lines.\n\nFor these transcriptions, see: Jean-Baptiste Camps, _La `Chanson d’Otinel’: édition complète du corpus manuscrit et prolégomènes à l’édition critique_, PhD thesis, dir. Dominique Boutet, Paris-Sorbonne, 2016,  DOI: [https://doi.org/10.5281/zenodo.1116735](10.5281/zenodo.1116735).\n\n\u003c!-- TODO: à compléter avec les autres manuscrits: Vatican, Mende, …  --\u003e\n\n## License\n\n\u003ca rel=\"license\" href=\"http://creativecommons.org/licenses/by/4.0/\"\u003e\u003cimg alt=\"Licence Creative Commons\" style=\"border-width:0\" src=\"https://i.creativecommons.org/l/by/4.0/88x31.png\" /\u003e\u003c/a\u003e\u003cbr /\u003eCette œuvre est mise à disposition selon les termes de la \u003ca rel=\"license\" href=\"http://creativecommons.org/licenses/by/4.0/\"\u003eLicence Creative Commons Attribution 4.0 International\u003c/a\u003e.\n\n## Contribute\n\nIf you want to contribute training data or models, you can do so by cloning the\nrepository and sending us a pull request, or by sending an email \nat _jbcamps at hotmail.com_ .\n\n## Cite this repository\n\nJean-Baptiste Camps (éd.), _FROC-MSS: Old French and Old Occitan Medieval Manuscripts HTR Data and Models_, Paris: École nationale des chartes (PSL), 2018, [https://github.com/Jean-Baptiste-Camps/FROC-MSS](https://github.com/Jean-Baptiste-Camps/FROC-MSS).\n\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FJean-Baptiste-Camps%2FFROC-MSS","html_url":"https://awesome.ecosyste.ms/projects/github.com%2FJean-Baptiste-Camps%2FFROC-MSS","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FJean-Baptiste-Camps%2FFROC-MSS/lists"}