Study of clustering algorithms for temporal segmentation of egocentric image sequences
Ver/ Abrir
Metadatos
Mostrar el registro completo del ítemcomunitat-uji-handle:10234/158176
comunitat-uji-handle2:10234/71345
comunitat-uji-handle3:10234/94547
comunitat-uji-handle4:
TFG-TFMEste recurso está restringido
Metadatos
Título
Study of clustering algorithms for temporal segmentation of egocentric image sequencesAutoría
Tutor/Supervisor; Universidad.Departamento
Traver Roig, Vicente Javier; Universitat Jaume I. Departament de Llenguatges i Sistemes InformàticsFecha de publicación
2018-07Editor
Universitat Jaume IResumen
In this project, we propose to explore an algorithm to perform an efficient low-level temporal
segmentation on egocentric image sequences. Specifically, the Warped K-Means (WKM) algorithm
was paid a special attention ... [+]
In this project, we propose to explore an algorithm to perform an efficient low-level temporal
segmentation on egocentric image sequences. Specifically, the Warped K-Means (WKM) algorithm
was paid a special attention as it is an adaptation to sequential data of the widely known k-means.
With the ultimate objective of first-person image sequences summarization in mind, this algorithm
is applied to segment the video into low-level events. After a preliminary analysis of the WKM, a
leader-based grouping algorithm is tested on the same data and for the same purpose, to evaluate
its interest for the problem at hand. Contrary to the WKM, the leader-based algorithm requires a
distance threshold to be specified instead of the number of clusters, which can be arguably easier
to set. The leader-based algorithm is first applied directly to our image sequences in order to group
frames corresponding to the same “event”, before being combined with the WKM to address this
problem. Experimental results over six image sequences from two different egocentric datasets
show the efficiency of the method for low-level segmentation. [-]
Palabras clave / Materias
Descripción
Treball final de Màster Universitari en Sistemes Intel.ligents (Pla de 2013). Codi: SIE043. Curs acadèmic 2017-2018
Tipo de documento
info:eu-repo/semantics/masterThesisDerechos de acceso
http://rightsstatements.org/vocab/CNE/1.0/
info:eu-repo/semantics/restrictedAccess
info:eu-repo/semantics/restrictedAccess