Pose2Room: Understanding 3D Scenes from Human Activities.

Yinyu Nie Angela Dai Xiaoguang Han Matthias Nießner

Published in: ECCV (27) (2022)

Keyphrases

human activities
d scene
human activity recognition
activity recognition
depth map
human computer interaction

scene understanding
single image
image based rendering
view dependent
sensor data
long video sequences

optical flow
pose estimation
temporal structure
realistic rendering
human actions

depth information
multi view
high quality
machine learning
action recognition