Course Description

If we want to develop intelligent agents that can interact, help and collaborate with humans, we need to endow said agents with the ability to perceive and understand humans from visual observations. These observations are not independent streams of pixels - they are a depiction of the underlying 3D structure. In this course, we will study the inference of the dynamic structure of humans from visual observations. We will cover a wide range of topics related to human body modeling, inference of 3D human structure and motion from images and video, as well as the use of these tools in related applications, e.g., in graphics, robotics, mixed reality, and biometrics. Most topics will be approached in the context of the recent advances in (deep) learning and the impact they have had in the progress of the area.

The format of this course will be a mix of lectures and student presentations/ seminar-style discussions. Students will be responsible for paper readings, class presentation, class participation, and completing a final project.

Online Platforms:

• Canvas for submitting paper reading reviews (latex template here) and course project materials

• Ed Discussion for online discussion

Prerequisites: The course assumes a good knowledge of computer vision and deep learning. We will not enforce any strict prerequisites, but intro courses to Computer Vision or Machine Learning would be very helpful.