Self-supervised depth super-resolution with contrastive multiview pre-training.
Xin QiaoChenyang GeChaoqiang ZhaoFabio TosiMatteo PoggiStefano MattocciaPublished in: Neural Networks (2023)
Keyphrases
- super resolution
- depth map
- low resolution
- high resolution
- image reconstruction
- low resolution images
- image restoration
- high resolution images
- motion blur
- super resolved
- motion estimation
- super resolution reconstruction
- resolution enhancement
- single frame
- image super resolution
- multi frame
- video super resolution
- multi view
- high quality
- adaptive regularization
- single image super resolution
- light field
- multiple views
- higher resolution
- back projection
- optical flow
- training set
- super resolution algorithm
- image patches
- d objects
- three dimensional
- computer vision