DIANOEMA: Visual analysis and sign recognition for GSL modelling and robot teleoperation

Here we present research work performed in the framework of the Greek national project DIANOEMA (GSRT, M3.3, id 35), focusing on the following activities: i) Development of innovative image analysis and computer vision algorithms for the effective visual analysis of video sequences, aiming at sign detection and tracking; ii) Creation of a video-corpus of the Greek Sign Language (GSL) and annotation and modelling of an indicative subset of it; iii) Automatic recognition of indicative categories of GSL gestures using automatic computer vision systems pre-trained on the GSL corpus, and combining AI techniques, machine learning and probabilistic analysis for the estimation of gesture instantiations. iv) Integration of the above into a pilot application system of robot tele-operation, on the basis of a pre-defined vocabulary of simple signs for the tele-operation control.