Christopher Kermorvant

I will be presenting how #Arkindex is being used to scale document processing to industrial scale for the #socface project @InedFr @PSEinfo @ArchivesnatFr

Very happy to be at @c2rmf with Mathieu Aubry @ImagineEnpc to present our platforms dedicated to AI applied to historical and cultural documents #AIKON #Arkindex @_Teklia_

@_Teklia_ @MuseeGuimet First presentation by Tayler Arnold @urichmond : Explainable and Auditable Search and Discovery of Visual Cultural Heritage Collections. Very inspiring overview of metadata, machine learning, deep learning, vLLMs for @DistantViewing

Christopher @kermorvant will be presenting TEKLIA's work on the analysis of old photographs from the @MuseeGuimet using open-source tools (#Arkindex, #YOLO, #Qwen, etc.)

Artificial Intelligence in Archives and Collections - Dealing with Images and Multimodal Cultural Heritage : an online and in-person conference at @LeibnizHI 12-13 December 2024 herder-institut.de/event/conferen…

@MissBrutus Félicitations ! Où peut-on trouver le manuscrit ?

Hier, après 3 ans de travail de recherche et la production d’un bon gros pavé de 400 pages, je soutenais ma thèse en humanités numériques, devant les membres de mon jury, une bonne partie de ma famille (présents ou à distance), des amis, et mes collègues passés et présents.

Le programme des deux jours ici : msh.uca.fr/content/journe…

Très heureux d'être à la @ClermontMsh pour deux jours consacrés à "Qu’est-ce que l’IA peut faire pour vous ? Journées IA pour les #SHS" Au programme, des LLM bien sûr, mais aussi #arkindex #machinelearning #opensource

@BAC_Chair @_BusinessArch @BusArchScot Too bad, next time !

@kermorvant @_BusinessArch @BusArchScot Alas no as I had to leave Glasgow today ☹️

Good morning from Glasgow! The obligatory run before ‘Liquid Gold’ the joint ⁦@_BusinessArch⁩ ⁦@BusArchScot⁩ which is taking place today.

Very pleased to be at Edrington/The Macallan in Glasgow to present our project on Richard @Hennessy 's correspondence exploration with #AI. Thanks to the @_BusinessArch /@BusArchScot for organizing this event @edringtoncomms bac.whitefuse.net/events/liquid-…

Très content de retrouver la promo 2025 du master #IODAA à @AgroParisTech pour une séance d'accompagnement professionnel sur l'évolution du métier de #datascientist : gestion de projet IA, l'impact des LLMs sur le métier, analyse de CV et entretiens individuels. @_Teklia_

@ed_hawkins @mikepriebe @ehsai1 @ed_hawkins a csv per page like this for you image rainfall_1903 ?

@mikepriebe @ehsai1 Yes please!

We now have around 350,000 images of daily rainfall observations taken between 1870 and 1960 from 1000s of sites around the UK. More than 100 million measurements. Very few are already digitised as usable data. This is a huge AI/ML challenge. Anyone want to help?

@ed_hawkins Hi @ed_hawkins , we have a model for data extraction in table that is currently being used to process the 30M pages of the French census. It has shown very good results on numbers in a previous project for the @ArchivesnatFr. We will be happy to help. Please contact me.

Oubliez Netflix, la série la plus passionnante de l'été est sur @lemonde : « 1940-1942, sur les traces du photographe inconnu » Un drame historique avec au générique @MRN_musee, les archives de @prefpolice, le SHD @DeHistorique, @ArolsenArchives et les archives du @Printemps

Solène Tarride, researcher at @_Teklia_ , will give a talk today at 6:00 PM EDT at #DH2024 on "A quantitative and qualitative evaluation of large scale handwriting recognition models for Norwegian", a joint project with @nasjonalbiblioteket

Christopher @kermorvant will give a talk today at 10:30 EDT during #DH2024 on "Semantic Content Description : How Contemporary Deep Learning Models Describe Early Japanese Photographs". A joint work with @MuseeGuimet

New version of the document-datasets.org portal with full description and bib refs for each dataset. The original paper by @koni_nik is now fully in the database! The next step is to allow public contributions. #ICDAR2024 @_Teklia_

@koni_nik also inspired by #HTRUnited by @Alix_Tz and @PonteIneptique

Based on an extensive survey paper by @koni_nik, we have released a first version of document-datasets.org , a portal for aggregating links to datasets for historical document processing. Feedbacks are welcome ! #ICDAR2024 #HTR

@Mila_Quebec Travailler au @Mila_Quebec en 2003 était déjà une expérience exceptionnelle, tant sur le plan scientifique qu'humain. Un souvenir toujours aussi vif et heureux !

Où sont les anciens de Mila aujourd'hui? rencontrez Christopher Kermorvant @kermorvant, ancien chercheur postdoc à Mila, aujourd'hui fondateur et directeur scientifique de Teklia. ow.ly/wA1P50Se9iO #MilaAlumni

@melissaterras @J_Nockels @pmgooding Very interesting paper @melissaterras, I fully agree with your recommendations and especially "Collaboration in tool and platform development". A step in this direction, we at @_Teklia_ have open-sourced our HTR platform #arkindex teklia.com/our-solutions/…

New paper! I'm proud of this one, with @J_Nockels & @pmgooding: How is HTR changing our information environment, & knowledge of the past? "The implications of handwritten text recognition for accessing the past at scale" #OpenAccess emerald.com/insight/conten…