Skip to main content
European Commission logo
Deutsch Deutsch
CORDIS - Forschungsergebnisse der EU
CORDIS
CORDIS Web 30th anniversary CORDIS Web 30th anniversary
Inhalt archiviert am 2024-05-07

Video indexing classification annotation and retrieval

Exploitable results

Within the framework of the ESPRIT project VICAR (Video Indexing Classification, Annotation and Retrieval) a non-linear video documentation and annotation software, the “VideoNavigator” has been developed. Relying fully on JAVA technology this program provides an elaborate set of functions for the annotation and logical structuring of digital video data. Recently, the “VideoNavigator” was selected as an innovative product during the international conference RIAO 2000. The topic of the conference which took place at the prestigious “College de France” in Paris in the mid of April was content-based multimedia indexing. The novel non-linear structuring of video contents allows for creating multiple annotation tracks, which may mutually overlap in time, thus offering the possibility of creating an arbitrary number of different annotation levels for one video. Various views allow for separated as well as for synoptic representation of the video annotations. In order to aid the annotation, the video navigator provides tools for automatic structuring of videos. In its current version the program automatically subdivides a video into shots and extracts key frames on a selectable granularity. Additional tools for extracting the camera motion and for finding people are also available. Manual annotation and structuring is thoroughly supported by an intuitive user interface. The annotation data footage can be searched using an integrated search mechanism that allows for textual queries. Additionally, the integrated search engine provides an associative retrieval option that allows for searching synonymous terms. With the query-by-image example similar frames and even video sequences can be retrieved. Currently two kinds of descriptors are employed, a colour and a texture oriented descriptor. The novel non-linear structuring of video contents allows for creating multiple annotation tracks that may mutually overlap in time, thus offering the possibility of creating an arbitrary number of different annotation levels for one video. Various views allow for separated as well as for synoptic representation of the video annotations. In order to aid the annotation, the video navigator provides tools for automatic structuring of videos. Additional tools for extracting the camera motion and for finding people are also available. Manual annotation and structuring is thoroughly supported by an intuitive user interface. The video navigator has numerous application scenarios. Its primary field of use are professional video archives as four of these institutions (ORF, SWR; SVT, NAA) directly took part in the project. The intention here is the in-depth documentation of valuable video material for archiving and retrieval purposes. However, the applicability of the video navigator is not restricted to large archives. Smaller archives or even individuals may structure and annotate their digital video footage with the VICAR software. As the software provides a standardized plug-in interface, it is easily extensible and may serve as a software basis for research projects on video analysis and structuring which is another promising field of exploitation.

Suche nach OpenAIRE-Daten ...

Bei der Suche nach OpenAIRE-Daten ist ein Fehler aufgetreten

Es liegen keine Ergebnisse vor