Visual/Video Processing
Contact persons: JP.Thiran at epfl.ch
An important element of multimodal multimedia information management systems is the development of unimodal primitives based on solid theoretical foundations and their correspondent integration into multimodal protocols. Our main focus in this framework corresponds to the analysis of visual information that includes digital and analog images, documents with different sort of semantic information and text. Specifically, the objectives of this IP is to develop cutting-edge algorithms for visual (images and videos) information processing, namely segmentation, feature extraction, representation, modelling and classification, with particular emphasis on face, gesture, people and handwriting image analysis, that are of central importance for IM2.
To achieve these objectives, we work in this IP will be divided in the following tasks:
Partners
ITS/EPFL, ASL/EPFL, Idiap, FKI/UniBE, BIWI/ETHZ.
