Multi-Frame Self-Supervised Depth with Transformers.
Vitor GuiziliniRares AmbrusDian ChenSergey ZakharovAdrien GaidonPublished in: CoRR (2022)
Keyphrases
- multi frame
- motion estimation
- point correspondences
- image sequences
- motion analysis
- super resolution
- single frame
- optical flow
- computationally efficient
- multiple frames
- filtering algorithm
- depth map
- optic flow
- super resolution image reconstruction
- temporal coherence
- stereo correspondence
- depth information
- video frames
- motion detection
- motion vectors
- viewpoint
- high quality