AI Improved Markerless Motion Capture For Dance

Motion capture is a technology to track human motion in 3D for replication with digital tools, such as mixed and virtual reality, video games, or movies. However, full 3D tracking requires the complex and finicky setup of multiple cameras that are prohibitively expensive to individuals or small studios. This is especially difficult with complex human movement such as in sports, dance, and more. While tracking with a single depth camera or webcam is possible, the quality of tracking is limited, and frequently tracked limbs will be lost due to fast 3D motion (e.g., a hand being obscured behind one’s back), which can ruin the capture, or require hours of tedious manual work to fix. Additionally, these cheaper capture methods are less stable, resulting in tracked motions appearing jittery, or jumping from position to position.

Faculty Supervisor:

Daniel Rea

Student:

Partner:

Spandrel Interactive Inc.

Discipline:

Computer science

Sector:

Professional, scientific and technical services

University:

University of New Brunswick

Program: