3D-LMNet: Latent Embedding Matching for Accurate and Diverse 3D Point Cloud Reconstruction from a Single Image.
Priyanka MandikalNavaneet K. L.Mayank AgarwalR. Venkatesh BabuPublished in: CoRR (2018)
Keyphrases
- single image
- shape recovery
- multiple images
- d scene
- super resolution
- light source
- image pairs
- feature matching
- indoor scenes
- image reconstruction
- high resolution
- shape from shading
- matching algorithm
- structured light
- partially occluded
- image correspondences
- intrinsic images
- high quality
- latent variables
- image matching
- computational complexity
- latent space
- multiscale