corpora studies on French Sign Language (LSF): elicitation, annotation, and analyse methods
Dicta-Sign project : constitution, annotation and analysis of a corpus of LSF dialog of 8 hours. Collaboration with IRIT lab for the constitution and with WebSourd company for the annotation. Annotation of linguistic structures. Constitution of a guideline for LSF annotation. The movies, annotations and guideline will be accessible for the scientific community and the end of the project.
Study on non-manual components: Emilie Chételat's PhD thesis (in French) on eyebrow movement and blinking; study on all the non-manual components within the Dicta-Sign project
Study on coarticulation : Jérémie Segouat's PhD thesis (in French) in collaboration with WebSourd company.
linguistic modelling : computing representation of the LSF functionning, semi-automatic translation between written French and LSF
Modelling of linguistic structures using Azalee formalism within the Dicta-Sign project
Project os semi-automatic translation from written French to LSF
virtual signer animation, based on LSF utterance generation and animation control.
Maxime Delorme's PhD thesis on realistic signs generation from Zebedee : first stage, generation of realistic postures.
Projects in collaborations
Dicta-Sign: European project which aim is to make communications more accessible to Deaf signers (Web 2.0).
WebSourd: Technology transfert and exploitation within a project of virtual signer for SNCF (French railway company)
Implementations and evaluations
SignStore : database of LSF lexical signs (almost 2000 entries). It will be put online during 2012.
OCTOPUS : software that concatenate predefined 3d animations in order to produce LSF sentences using a virtual signer. Used by the WebSourd company for the SNCF company (French railway company) in order to generate LSF message in railway stations in France.
DIVA : framework for 2d1/2 animation of a virtual signer on the web, in collaboration with Limsi's AMI group. Example of prototype.
GeneALS : framework for automatic generation of LSF utterances