Description of Research Activities

The group is interested in the analysis of visual content, in particular spatio-temporal, such as video or even multi-modal data: sensor signals, video, images. Today we develop deep learning AI tools for the analysis of these contents, the particular interest is given to self-attention models in deep networks and transformers, their evaluations in relation to psychovisual data. As part of the transverse axis "AI" of LABRI, we develop tools for explanations of AI decisions in information classification problems.

In motion estimation we use classical approaches by optical flow. In classification we found our research on machine learning -Deep Learning with NNs.