Depth Based Semantic Scene Completion with Position Importance Aware Loss.
Yu LiuJie LiXia YuanChunxia ZhaoRoland SiegwartIan ReidCesar CadenaPublished in: CoRR (2020)
Keyphrases
- defocused images
- d scene
- depth estimation
- semantic context
- depth map
- real scenes
- scene geometry
- relative position
- depth cues
- depth ordering
- scene understanding
- image sequences
- video sequences
- single image
- imaging sensors
- depth images
- positional information
- natural language
- three dimensional
- precise localization
- dynamic scenes
- depth information
- semantic information
- high level
- semantic web
- semantic similarity
- input image
- stereo vision
- scene images
- low level features
- position and orientation
- stereoscopic images
- moving objects
- scene structure
- scene classification
- relative depth
- multiple images
- single shot
- depth discontinuities
- image based rendering
- view synthesis
- spatial location
- ego motion
- complex scenes