Learning Implicit Representations for Taking Head Generation
Project Description
In this project, we will study an interesting computer vision problem (i.e. talking head generation), which is popular and widely used in various virtual/augmented reality application scenarios nowadays. More specifically, we will investigate how to learn the novel and effective implicit representations, and apply them for generating high-quality talking head videos.
Supervisor
XU Dan
Quota
3
Course type
UROP1000
UROP1100
UROP2100
Applicant's Roles
Study and investigate the literature and conduct research and development on the objectives.
Applicant's Learning Objectives
- Investigate the state-of-the-arts in the field of video-based talking head generation
- Study on the implicit representation learning
- Combing implicit representations with generative networks for high-fidelity talking-head generation
- Study on the implicit representation learning
- Combing implicit representations with generative networks for high-fidelity talking-head generation
Complexity of the project
Moderate