MOFT: Monocular odometry based on deep depth and careful feature selection and tracking.
Karlo KoledicIgor CvisicIvan MarkovicIvan PetrovicPublished in: ICRA (2023)
Keyphrases
- feature selection
- model based tracking
- visual odometry
- depth cues
- position estimation
- human body tracking
- real time
- depth images
- visual tracking
- image sequences
- ego motion
- particle filter
- depth perception
- pose tracking
- machine learning
- monocular images
- particle filtering
- depth information
- object tracking
- text categorization
- pose estimation
- depth map
- feature set
- autonomous navigation
- feature extraction
- kalman filter
- appearance model
- motion segmentation
- feature selection algorithms
- model selection
- support vector
- optical flow
- mean shift
- visual slam
- camera motion
- pose recovery
- classification accuracy
- support vector machine
- computer vision
- mutual information
- simultaneous localization and mapping
- depth data
- robust tracking
- human pose
- text classification
- field of view
- stereo vision
- motion model