Moments of Inertia

These types of automated video analysis does the TIB AV-Portal use:

Scene recognition — shot boundary detection segments the video on the basis of image characteristics. A visual table of contents created from this provides a quick overview of the video content, facilitating accurate access.

Text recognition – intelligent character recognition determines and indexes written language (such as text written on slides), and makes it searchable.

Speech recognition – speech to text notes down spoken language in the video in the form of a transcript, which is searchable.

Image recognition – visual concept detection indexes the moving image using subject-specific and multidisciplinary visual concepts (for example, landscape, façade detail, technical drawing, computer animation or lecture).

Indexing – named-entity recognition describes individual video segments with semantically associated terms. This ensures that synonyms or hyponyms of any search terms entered are also searched for automatically, increasing the number of hits.