S3E-GNN: Sparse Spatial Scene Embedding with Graph Neural Networks for Camera Relocalization.
Ran ChengXinyu JiangYuan ChenLige LiuTao SunPublished in: CoRR (2022)
Keyphrases
- neural network
- real scenes
- multiple images
- camera images
- relative position
- scene structure
- camera pose
- scene geometry
- autocalibration
- d scene
- monocular slam
- ground plane
- visual scene
- acquired images
- fish eye
- graph embedding
- stereo camera
- camera tracking
- active illumination
- moving camera
- vision system
- camera parameters
- uncalibrated cameras
- video sequences
- spectral signatures
- virtual camera
- spatial location
- camera views
- single camera
- spatial relations
- field of view
- camera motion
- spatial information
- camera positions
- structure from motion
- scene reconstruction
- white balance
- object motion
- temporal resolution
- ego motion
- live video
- imaging process
- line features
- high dimensional
- camera movement
- image sequences
- single image
- dynamic scenes
- hand held
- multiple cameras
- camera calibration
- point correspondences
- feature points
- surveillance system
- defocused images
- optical axis
- depth map
- moving objects
- depth estimation
- sparse representation
- vanishing points
- position and orientation
- multi camera
- video camera
- dimensionality reduction
- space time