CSGAT-Net: a conditional pedestrian trajectory prediction network based on scene semantic maps and spatiotemporal graph attention.
Xin YangJiangfeng FanXiangcheng WangTao LiPublished in: Neural Comput. Appl. (2024)
Keyphrases
- moving objects
- prediction accuracy
- crowded scenes
- object trajectories
- moving points
- spatial and temporal
- graph model
- semantic context
- three dimensional
- graph representation
- d scene
- trajectory data
- graph structure
- space time
- spatio temporal
- single image
- random walk
- scene understanding
- semantic web
- image sequences
- object detection
- real scenes
- semantic similarity
- dynamic scenes
- input image
- natural language
- semantic information
- complex scenes
- prediction error
- graph theory
- video surveillance
- directed graph
- location prediction
- continuously moving
- graphical structure
- video scene
- semantic video retrieval
- topological map
- scene classification
- motion patterns
- multiple images
- pedestrian detection
- low level features
- visual attention
- graphical models
- video sequences
- high level