Randomly connected neural networks for self-supervised monocular depth estimation.
Samyakh TukraStamatia GiannarouPublished in: Comput. methods Biomech. Biomed. Eng. Imaging Vis. (2022)
Keyphrases
- depth estimation
- monocular images
- depth cues
- stereo vision
- depth map
- depth perception
- stereo matching
- depth information
- dynamic scenes
- image sequences
- depth estimates
- model based pose estimation
- real scenes
- stereo pair
- scene understanding
- disparity map
- super resolution
- multi view
- optical flow
- pose estimation
- stereo images
- input image
- vision system
- specular reflection
- high quality
- three dimensional
- depth from defocus
- preprocessing
- feature matching
- background subtraction
- object recognition
- ego motion
- face recognition
- d scene
- matching algorithm