A longstanding challenge in computer vision is object permanence, a well-known concept in psychology that involves understanding that the existence of an object is separate from whether it is visible at any moment.

To tackle this issue, Columbia Engineering researchers taught neural networks to anticipate what scenes look like in the future by watching many videos.

Given a monocular video as input, the framework produces a 4D representation that captures the entire scene content along with all the static and dynamic objects within it over time.

Read the summary: “Revealing Occlusions with 4D Neural Fields”
https://occlusions.cs.columbia.edu

Read the full story on our website: engineering.columbia.edu

Video by Basile Van Hoorick