Tracking people and objects in 3D


At a glance

We propose extending our work on tracking humans in 3D to also track objects via spatial-temporal 3D reconstruction. We pose the problem as finding the 3D shape, pose, location, and appearance, of multiple people and objects, and track changes over time in an input video. Objects here could include cars, bicycles etc.

principal investigatorsresearchersthemes

Jitendra Malik

Angjoo Kanazawa


tracking, shape reconstruction, people