DCU-NET: Self-supervised monocular depth estimation based on densely connected U-shaped convolutional neural networks.
Qiumei ZhengTao YuFenghua WangPublished in: Comput. Graph. (2023)
Keyphrases
- depth estimation
- convolutional neural networks
- monocular images
- densely connected
- depth cues
- stereo vision
- depth map
- stereo matching
- depth estimates
- depth information
- dynamic scenes
- real scenes
- scene understanding
- stereo pair
- image sequences
- graph mining
- pose estimation
- weighted graph
- super resolution
- feature matching
- community structure
- information networks
- belief propagation
- high resolution
- vision system
- community detection
- multi view
- specular reflection
- d objects
- disparity map
- computer vision
- object recognition
- high quality
- three dimensional