Human Pose and Action Recognition using Negative Space Analysis

Human Pose and Action Recognition using Negative Space Analysis

This thesis proposes a novel approach to extracting pose information from image sequences. Current state of the art techniques focus exclusively on the image space occupied by the body for pose and action recognition. The method proposed here, however, focuses on the negative spaces: the areas surro...

Full description

Bibliographic Details
Main Author: Janse Van Vuuren, Michaella
Other Authors: Electrical Engineering
Format: Others
Language:en
Published: University of Cape Town 2016
Subjects:
Image processing
pattern recognition
computer vision
human pose and action recognition using negative space analysis
surveillance
chroma-key
silhouette
preprocessing
color coded
colour-coded
horizontal scanning
vertical scanning
RGB colour image
bounding box
multi-view
multi view
aerobics
exercise
feature extraction
clustering
over generalisation
over fitting
k-means
SOM
self organising map
automatic partitioning
pose labelling
pose clusters
pose classification
correlation
sequence
feedback
recognising actions
visualization
feature plots
direct able
directable character
database
animated sequences .
QA75
NX
QA76
TA
T1
TK
Q1
Online Access:http://hdl.handle.net/10919/71571

Internet

http://hdl.handle.net/10919/71571

Similar Items