MOFO: MOtion FOcused Self-Supervision for Video Understanding.
Mona AhmadianFrank GuerinAndrew GilbertPublished in: CoRR (2023)
Keyphrases
- space time
- spatial and temporal
- input video
- video sequences
- temporal filtering
- video data
- key frames
- moving camera
- successive frames
- video representation
- low frame rate
- motion features
- object motion
- motion analysis
- visual cues
- visual data
- temporal consistency
- motion estimation
- video objects
- optical flow
- multimedia
- image sequences
- periodic motion
- video frames
- shot change detection
- motion detection
- temporal continuity
- motion trajectories
- temporal domain
- object detection and tracking
- motion patterns
- static images
- video signals
- video scene
- temporal structure
- surveillance videos
- video analysis
- motion segmentation
- human motion
- video content
- multimedia data
- video streams
- camera motion
- moving objects
- real time
- motion capture data
- layered representation
- wide area motion imagery
- single frame
- video summarization
- image frames
- dynamic textures
- dynamic scenes
- spatio temporal
- reference frame
- humanoid robot
- motion parameters
- stationary camera
- recognizing human actions
- motion compensation
- motion model
- human body
- computer vision