Pix2NeRF: Unsupervised Conditional π-GAN for Single Image to Neural Radiance Fields Translation.
Shengqu CaiAnton ObukhovDengxin DaiLuc Van GoolPublished in: CoRR (2022)
Keyphrases
- single image
- shape recovery
- multiple images
- image pairs
- light source
- spherical harmonics
- d scene
- semi supervised
- super resolution
- partially occluded
- illumination conditions
- three dimensional
- indoor scenes
- scene geometry
- structured light
- shape from shading
- vanishing points
- lighting conditions
- cast shadows
- manhattan world
- structuring elements
- specular reflection
- viewpoint
- computer vision
- fundamental problem in computer vision