HumanActivity Description UsingMotion Verbsin
Miyoung Cho, Pankoo Kim · 2006
pkkimgchosun.ac.kr Abstract- Tobridge thegapbetween computational models for activity detection systems havebeendeveloped inthelast spatio-temporal relations andpeople's useofmotion verbsin decades. Through researching abouttheformer workswe their natural language, amodelforthespatio-temporal relations found that wehavetospend alotoftimeonexplaining and wascalibrated formotion verbs. So,inthespecific domain, we annotating thehumanaction evensuchasatiny action: raising introduce recognition ofhumanaction/activity fromvideos usingarm,moving theheadorassort. Notevenreferring tothe motion verbs onontology ofWordNet. event, we canimage howittime andlabor consuming for Keywords-Motion verbs, WordNet,Spatio-temporal analyzing. Inorder toimprove thesystem efficiency, inthis Keywiors - paper, weaimtodefine anevent ontology that allows natural