Title Žmogaus pozų atpažinimo algoritmo ir jį realizuojančios programinės įrangos sukūrimas bei algoritmo tikslumo tyrimas /
Translation of Title Implementation of human pose detection algorithm and its precision research.
Authors Ryselis, Karolis
Full Text Download
Pages 91
Keywords [eng] hausdorff distance ; pattern matching ; kinect ; yoga
Abstract [eng] Healthy life style is getting more and more popular. More and more people embrace it. One of the main activities contributing towards this life style is yoga. It improves human’s physical and mental health. Nowadays everyone can practice yoga, but assistance from a professional trainer is required to do it correctly. The assistance could be provided in a yoga center. People who live in remote areas or work at the same time as yoga centers cannot take part in those yoga trainings. They lose the ability to get professional yoga training. Yoga could be done incorrectly without a trainer and therefore it is only at amateur level. People want to practice yoga at home as well. A professional trainer is not available there. The goal of this work is to create a motion tracking component for a virtual yoga trainer. The main piece of motion tracking component is motion tracking algorithm. It is based on principles of Hausdorff distance and Procrustes analysis and compares two sequences of human poses efficiently. Assumptions based on yoga applications are made in this work. They allow optimizing existing algorithms, applying approximate methods that fit well to measure metrics for human movements. Software that implements these algorithms is created in this work and its efficiency and precision are assessed, algorithm implementation is compared to other existing solutions. The motion tracking component could be applied in other areas – health promotion, rehabilitation and other areas that require correct completion of a physical action.
Dissertation Institution Kauno technologijos universitetas.
Type Master thesis
Language Lithuanian
Publication date 2017