Home Page

Research activities description

The developed research works are focused on the problem of multimedia information for services and medical imaging. This theme is involved in notional research programs, such as: GDR-ISIS in the Theme B - Axis 3 Image and Vision, involved in the consortium IRIM for the campaign TRECVID. The main research axes are:

AIV
Copyright: CERIMES & Corpus LaBRI

The developped methodological tools represent motion estimation at pixel level (optical flow) and at region/object level (appearant affine motion models) Interaction of these tools and classical opproach (morphologicalare, statistical) are studied to extracti the object of interest in video sequences. The combination of temporal and spatial approaches allows to achieve:

  • Robust object segmentation of video streams
  • Extraction of descriptors with colorimetric and dynamic propoerties from audio-visual contents toward the MPEG7 standard, and the indexation of content for general purpose navigation

Large Egocentric Gaze Objects Dataset

Access the dataset and its description

Key words

Motion Analysis and Indexation, Spatio-Temporal based Segmentation of SD and HD, Video Sequence Segmentation in Cuts and Scenes, Indexation of Compressed Multimedia Streams, Low-Level Video Descriptors Extraction for Video Indexing, Video Survailliance Applications, Biomedical Imaging Motion Estimation.