Deep Unsupervised 3D SfM Face Reconstruction Based on Massive Landmark Bundle Adjustment.

Yuxing Wang Yawen Lu Zhihua Xie Guoyu Lu

Published in: ACM Multimedia (2021)

Keyphrases

bundle adjustment
structure from motion
scene reconstruction
reprojection error
piecewise planar
multi view
multi view stereo
point correspondences
camera motion
least squares

projective reconstruction
camera parameters
nonlinear optimization
scene structure
missing data
levenberg marquardt
feature correspondences
camera pose
image sequences
image correspondences

pinhole camera
semi supervised
feature tracking
face images
facial expressions
machine learning
stereo rig
camera positions
autocalibration
computer vision

maximum likelihood
real time
single view
three dimensional
active learning
linear combination
depth map
motion parameters
focal length