Estimating more camera poses for ego-centric videos is essential for VQ3D.
Jinjie MaiChen ZhaoAbdullah HamdiSilvio GiancolaBernard GhanemPublished in: CoRR (2022)
Keyphrases
- ego centric
- hand detection
- camera pose
- structure from motion
- camera motion
- video sequences
- bundle adjustment
- point correspondences
- camera parameters
- pixel level
- skin color
- video frames
- dynamic scenes
- position and orientation
- focal length
- object manipulation
- video data
- image sequences
- camera calibration
- multi view
- urban environments
- moving camera
- line segments