Linguistic Techniques for Event Recognition
Lists
Only you see this
Abstract
Linguistic techniques for event recognition use elements of language such as vocabulary, grammar, expressions, syntax, and semantics to represent and recognize events. Visual event recognition systems use linguistic techniques to describe complex events by the composition of primitive sub-events according certain rules (spatial, temporal, logical, etc.). This allows systems to represent a large set of events while modeling a much smaller set of primitive events by low-level processing. Representational and computational properties of these systems depend on the types of allowable relationships between sub-events. Some systems also aim for representations that are closer to natural language to facilitate user interaction, i.e., to describe events, make queries, or interpret results. Hidden Markov Models (HMMs) [10] are statistical Markov models that decompose temporal sequences into discrete time steps, each with an associated unobserved...
BibTeX
@inproceedings{morariu2014linguistic,
author = {Vlad I. Morariu},
title = {{Linguistic Techniques for Event Recognition}},
booktitle = {Computer Vision, A Reference Guide},
year = {2014},
url = {https://openreview.net/forum?id=McnisUC1Ns},
}Loading related content...