Bilkent University
Department of Computer Engineering

---

CS 590/690 SEMINAR

 

Personality Style Transfer Between Animations with Deep Learning

 

Arçin Ülkü Ergüzen

Master Student
(Supervisor: Prof.Dr.Uğur Güdükbay)
Computer Engineering Department
Bilkent University

Abstract: Humans, being social creatures, are always in contact with each other and the environment. While providing this communication, speech, gestures, facial expressions, and body movements play an active role, while at the same time, they reflect the personalities of the individuals. And today, conveying personality well is crucial to contributing more realistic appearance of animations, video game characters, virtual assistants, and environments we see on screens. My aim in this project is to successfully transfer motion from one animation to another based on the personalities extracted from the motion with the help of Deep Learning. Currently, there exist multiple motion style transfer research but none of them applies any personality information to the generated motion. Also, in the later stages of my thesis, I want to construct a system to directly alter an animation based on the learnt personality weights.

 

DATE: April 17, Monday @ 15:50 Place: Zoom