Electronic Theses and Dissertations

Identifier

6027

Date

2017

Document Type

Thesis

Degree Name

Master of Science

Major

Electrical and Computer Engr

Concentration

Computer Engineering

Committee Chair

Bonny Banerjee

Committee Member

Madhusudhanan Balasubramanian

Committee Member

Aaron L Robinson

Abstract

Realistic synthesis of human actions is a challenging problem. This thesis investigates the problem of synthesizing actions, with individual variability, under different emotions. Current actions/gesture synthesis, understanding and recognition models do not provide a general framework for synthesizing an extensive range of actions over a large range of emotions.The literature on spectral style transfer provides a plethora of viable approaches for transferring the style of action learned from one individual to another. Our idea is to consider an emotion as a style then use a style transfer algorithm for transferring an emotion from one action to another. This allows us to synthesize any action over a large range of emotions. Experiments reported in this thesis are based on genarating18 actions with five emotions using the Kinect skeleton. The quality of the synthesized actions over time is evaluated through a subjective perception test, which is a standard in the domain of gesture synthesis.

Comments

Data is provided by the student.

Library Comment

Dissertation or thesis originally submitted to the local University of Memphis Electronic Theses & dissertation (ETD) Repository.

Download

Share

COinS