Computer analysis of images and videos is a major area of computer science. Advances are needed in computer understanding and organization of such “visual content” to improve the way we humans interact with and search through large catalogues of videos.
To this end, we develop new representations of video content which allow determination of relevant constituents within a video, together with automatic generation of textual descriptors.
Medical imaging raises specific challenges: there we may be looking for tumors, heart conditions, cognitive disorders,… based on data such as eg functional MRI. To address these challenges we develop efficient detectors, personalized to the patient under study, leveraging in particular so-called random decision forests.