|
Managing large volumes of video data poses several
significant technical challenges. The first level of challenges is in
compression, streaming, and editing of video. The next challenge is in
content-based retrieval of video. This requires that the video be indexed using
several modalities, including the image sequence, the audio/speech track, and
closed captioning if available. The image sequence indexing can be based on
several criteria, beginning with the location of visual scene changes,
localizing human faces in the video, performing analysis of the camera and
object motion, etc. Tasks in the audio domain include speaker change detection,
speaker identification, speech indexing, etc.
|
