SOSD-Net: Joint semantic object segmentation and depth estimation from monocular images.
Lei HeJiwen LuGuanghui WangShiyu SongJie ZhouPublished in: Neurocomputing (2021)
Keyphrases
- monocular images
- object segmentation
- depth estimation
- stereo vision
- object detection
- scene understanding
- depth map
- image segmentation
- graph cuts
- foreground objects
- stereo matching
- dynamic scenes
- depth information
- segmentation method
- shape prior
- segmentation algorithm
- human body
- real scenes
- object categories
- pose estimation
- stereo pair
- image sequences
- super resolution
- disparity map
- feature matching
- d scene
- specular reflection
- three dimensional
- markov random field
- human pose