Multimedia information, and particularly video, is becoming critical for the content industry, particularly in TV industry, but also for other domain such as technical or professional training. To cope with these needs, we propose to push the state-of -the-art of video annotation and retrieval systems. The work has been split into several sections: requirements of end-users in two applications domains, technologies focused on the basic technologies needed for analysing, coding, indexing, archiving and retrieval of videos, system focused on a video studio and a distributed system for the video retrieving, applications experiments to tests the results. (One for TV and one for professional training). At the end of the project we will prepare exploitation and dissemination of the results.
Handling rich content multimedia information, and particularly video, is becoming critical for the content industry. The multiplication of cable and satellite channels and the possibility of multiplexing the digital images turn the pressure towards the content industry. To cope with these needs, the systems aimed consist in content preparation workshops or studios to be integrated into the end-users workflow allowing to index raw footage prior to archiving and retrieve video material by iterative and single level operator queries. We propose to push the state-of-the art of video annotation and retrieval systems to the extent that the proposed system will (semi) automatically extract formal information from videos, like shot boundaries, camera movements, and scene clusters;
The work has been split into four major sections: requirements, technologies, system and applications.
The requirements section will propose a list of system requirements and user requirements defined by several end-users in different application contexts. It will offer the possibility to involve end-users very early in the process of the definition of the solution.
The technological section is focused on the basic technologies needed for analysing, coding, indexing, archiving and retrieval of videos decomposed into three work packages. They will not be developed from scratch, but in fact will be based on existing prototypes. So the efforts will be essentially devoted to enhancement, complementation, and integration of existing technologies brought by the participants.
The system section has been focused on the part of the system that will cope with all the user needs: firstly a video studio for video, indexing preparation, and secondly major exploitation components needed for video retrieving in two versions: light equipment for small end-users and high end equipment for service providers or larger end-users. These efforts are also split into three workpackages.
The application section will be devoted to the tests of the results. Two sets of applications have been selected to be sure of the generality of the approach. One will be linked to TV film producing and exploitation, the second one will be devoted to prepare exploitation and dissemination of the results. So this section will comprise also three workpackages.
Main reviews at T0+6, T0+12, T0+18, Final review at T0+24 Technologies prototypes and documents 1st version T0+7, 2nd version T0+12 Systems prototypes and documents 1st version T0+9, 2nd version T0+14 Applications and experiments (TV and training) T0+22 Results exploitation and market access T0+24.
Funding SchemeCSC - Cost-sharing contracts