Database Reference
In-Depth Information
Fig. 1.1
(
a
) User-controlled relevance feedback system. (
b
) Pseudo-relevance feedback system
an input video via an orderly sequential process. Initially, domain knowledge
independent local descriptors are extracted homogeneously from the input
video sequence. The video's genre is identified by applying k-nearest neighbor
(k-NN) classifiers onto the obtained video representation, with various dissimilarity
measures assessed and evaluated analytically. Subsequently, an unsupervised
probabilistic latent semantic analysis (PLSA) based algorithm is employed on
the same histogram-based video representation to characterize each frame of video
sequence into one of the representative view groups. Finally, a hidden conditional
random field (HCRF) structured prediction model is utilized for detecting events of
interest. In a trial evaluation, sports videos were used.
Search WWH ::
Custom Search