Motionsnap: A Motion Sensor-Based Approach for Automatic Capture and Editing of Photos and Videos on Smartphones.
Adil KarjauvSanzhar BakhtiyarovChaoning ZhangJean-Charles BazinIn So KweonPublished in: ICME (2021)
Keyphrases
- motion features
- moving camera
- video footage
- video sequences
- motion estimation
- key frames
- camera motion
- periodic motion
- sensor data
- video content
- motion analysis
- human motion
- static images
- input video
- image sequences
- video frames
- dynamic scenes
- visual features
- successive frames
- recognition of human actions
- temporal coherence
- optical flow
- mobile devices
- video stabilization
- video data
- motion parameters
- motion capture data
- motion capture
- motion model
- high frame rate
- space time
- photo collections
- motion cues
- video representation
- global motion
- sensor networks
- spatio temporal
- moving objects
- video summarization
- video objects
- motion trajectories
- motion segmentation
- video clips
- fully automatic
- motion patterns
- spatial and temporal
- human activities
- recognizing human actions
- robot motion
- daily life
- video analysis
- adjacent frames
- visual data
- dynamic textures
- temporal domain
- action recognition