Segment-Tube: Spatio-Temporal Action Localization in Untrimmed Videos with Per-Frame Segmentation.
Le WangXuhuan DuanQilin ZhangZhenxing NiuGang HuaNanning ZhengPublished in: Sensors (2018)
Keyphrases
- spatio temporal
- temporal segmentation
- human actions
- fully unsupervised
- video frames
- spatial and temporal
- optimal segmentation
- video objects
- video sequences
- action recognition
- accurate segmentation
- view invariant
- spatio temporal patterns
- action classification
- optic disc
- temporal coherence
- space time
- recognition of human actions
- video representation
- key frames
- input video
- temporal domain
- neighboring frames
- accurate localization
- single frame
- rigidly moving
- segmentation algorithm
- moving objects
- image sequences
- human activities
- automatically segmented
- level set
- figure ground
- segmentation scheme
- human motion
- video scene
- successive frames
- automatic segmentation
- spatio temporal data
- video database
- image segmentation
- image segments
- medical images
- recognizing human actions
- video content
- segmentation method
- activity recognition
- foreground regions
- shape prior
- video analysis
- camera motion
- adjacent frames
- video data
- segmented regions
- object segmentation
- spatio temporally
- homogeneous regions