Multi-Agent Embodied Visual Semantic Navigation with Scene Prior Knowledge.
Xinzhu LiuDi GuoHuaping LiuFuchun SunPublished in: CoRR (2021)
Keyphrases
- prior knowledge
- multi agent
- semantic context
- observed scene
- visual data
- visual information
- scene categorization
- d scene
- visual scene
- visual navigation
- semantic content
- multiple images
- single image
- scene understanding
- semantic information
- visual features
- visual context
- visual appearance
- visual concepts
- image sequences
- scene categories
- intelligent agents
- semantic space
- low level
- scene images
- video sequences
- high level
- visual input
- spatial layout
- visual environment
- real scenes
- semantic similarity
- low level features
- three dimensional
- prior information
- reinforcement learning
- domain knowledge
- image browsing
- spatial relations
- training data
- multi agent systems
- complex scenes
- input image
- moving objects
- video data
- natural language
- high level semantics
- depth map
- multiagent systems
- image regions
- scene recognition
- human vision
- indoor environments