Ontology-based human behavior indexing with multimodal video data

Synthesizing Event-Centric Knowledge Graphs of Daily Activities Using Virtual Space

“…Oh et al [11] detected events and then constructed KGs based on the data collected from mobile phones and wearable devices and video data pertaining to the behavior of older adults living in nursing care facilities. Vizcarra et al [12] constructed KGs based on the object recognition and manual annotation results of video data capturing the behavior of older adults living in nursing homes.…”

“…Thus, we referred only to the ontology design pattern, in which the event classes were directly connected to an agent, a location, and time. Our original classes and properties, which were similar in usage to the SEM and EO, were linked to the corresponding classes and properties using skos:closeMatch 12 .…”

“…[6] To date, daily activity data are typically collected from physical locations, such as care facilities for older adults [7], smart homes [8], and experimental facilities [9], [10] that imitate living homes. Although some previous studies [11], [12] converted the collected data into KGs, these studies required the use of physical equipment, experimental facilities, and human subjects. Representing a physical environment in virtual space and executing various simulations allow us to freely accumulate the data required for analyzing daily life and constructing EQA/VQA prediction models.…”

“…http://motools.sourceforge.net/event/event.html12 https://www.w3.org/2004/02/skos/ 13 https://www.w3.org/TR/turtle/…”

OntoImVid-ontology based semantic image/video processing

“…Vizcarra J et al [4] elaborated a methodology aimed at representing and indexing the semantics of human behaviors recorded in video data. The contribution was able to generate ontology-based RDF knowledge graphs with explicit knowledge (ontologies and knowledge bases).…”