MGDepth: Motion-Guided Cost Volume For Self-Supervised Monocular Depth In Dynamic Scenarios.
Kaichen ZhouJia-Xing ZhongJia-Wang BianQian XieJian-Qing ZhengNiki TrigoniAndrew MarkhamPublished in: CoRR (2023)
Keyphrases
- image sequences
- depth cues
- optical flow
- ego motion
- motion estimation
- motion analysis
- relative depth
- depth perception
- space time
- camera motion
- motion model
- stereo vision
- motion detection
- multiple visual cues
- motion tracking
- depth information
- perspective images
- visual odometry
- dynamic environments
- motion segmentation
- visual cues
- point correspondences
- depth map
- apparent motion
- feature points
- particle filter
- human body tracking
- pose estimation
- motion estimates
- motion capture
- motion field
- motion parameters