Human Motion Synthesis

A long-standing goal in computer vision and graphics is to capture, model, and realistically synthesize human behavior. In this seminar, we aim to study in detail the field of 3D human motion generation.

This course is suitable for Bachelor and Master level students with the solid knowledge on the following prerequisites:

• Knowledge of linear algebra and calculus, numerical optimization, probability theory.
• Prior participation in one of: Deep Learning, Probabilistic Machine Learning is recommended.

Schedule for the Seminars

Seminar	Date	Time	Title
Session 01	October 27, 2023	9:00-10:00	Introduction
Session 02	November 3, 2023	9:00-13:00	Scene-agnostic motion synthesis and static humans in the scene
Session 03	November 24, 2023	9:00-13:00	Synthesis in complex interactions
Session 04	December 1, 2023	9:00-13:00	Text-to-motion and motion diffusion models
Session 05	December 8, 2023	9:00-13:00	Motion generation and its applications

• Presentation/ Attendance
• Active participation in all sessions.
• Short presentation on a selected topic/paper: 10 minutes talk + 10 minutes questions
• Long presentation on a selected topic/paper: 20 minutes talk + 20 minutes questions
• Read paper, which is to be presented and submit at least 2 questions before the seminar to one of our TAs (below)

• Grading scheme:
• Students are graded based on active participation(~40%) of the sessions and presentations(~60%).
• Prior questions count towards part of the credit for participation.

• Please register at ILIAS.
• Note that the number of available presentation slots is limited to 10. These slots will be allocated in a first-come, first-serve basis.

Xiaohan Zhang (mail).