Exploring Optical-Flow-Guided Motion and Detection-Based Appearance for Temporal Sentence Grounding.
Daizong LiuXiang FangWei HuPan ZhouPublished in: CoRR (2022)
Keyphrases
- optical flow
- dynamic textures
- image sequences
- motion model
- motion field
- optical flow estimation
- motion estimation
- moving objects
- motion parameters
- binocular disparity
- spatial and temporal
- camera motion
- spatio temporal
- optical flow computation
- motion segmentation
- space time
- image motion
- optical flow fields
- detection algorithm
- feature tracking
- adjacent frames
- object detection
- temporal consistency
- computer vision
- velocity field
- temporal segmentation
- dense optical flow
- object motion
- consecutive frames
- moving camera
- flow field
- vector field
- motion trajectories
- static images
- natural language
- motion patterns
- motion analysis
- optical flow algorithm
- visual motion
- smoothness constraint
- motion discontinuities
- rigid body motion
- appearance features
- motion estimates
- motion cues
- motion features
- ego motion
- appearance model
- d scene