Scale Recovery for Monocular Visual Odometry Using Depth Estimated with Deep Convolutional Neural Fields.
Xiaochuan YinXiangwei WangXiaoguo DuQijun ChenPublished in: ICCV (2017)
Keyphrases
- visual odometry
- depth images
- ego motion
- network architecture
- autonomous navigation
- long range
- depth information
- field of view
- neural network
- kalman filtering
- image sequences
- depth map
- simultaneous localization and mapping
- pose estimation
- computer vision
- real time
- position information
- estimation process
- d scene
- stereo camera
- scale space
- low resolution
- dynamic scenes
- optical flow
- range data
- camera motion
- multi camera
- vision system
- high resolution
- robust estimation
- motion field
- camera pose
- single image