Transformers in Unsupervised Structure-from-Motion.
Hemang ChawlaArnav VarmaElahe AraniBahram ZonoozPublished in: CoRR (2023)
Keyphrases
- structure from motion
- bundle adjustment
- camera parameters
- image sequences
- computer vision
- missing data
- point cloud
- feature tracking
- scene structure
- reconstruction from multiple images
- camera motion
- semi supervised
- relative pose
- image based localization
- camera positions
- camera pose
- focal length
- feature correspondences
- reprojection error
- multibody
- factorization method
- video sequences