TAFormer: A Unified Target-Aware Transformer for Video and Motion Joint Prediction in Aerial Scenes.
Liangyu XuWanxuan LuHongfeng YuYongqiang MaoHanbo BiChenglong LiuXian SunKun FuPublished in: CoRR (2024)
Keyphrases
- dynamic scenes
- space time
- video footage
- video scene
- dynamic textures
- spatial and temporal
- moving camera
- object motion
- temporal filtering
- video sequences
- input video
- video frames
- video data
- motion estimation
- visual cues
- low frame rate
- key frames
- static images
- dynamic background
- motion features
- robust tracking
- surveillance videos
- moving target
- visual data
- video content
- successive frames
- temporal continuity
- motion analysis
- motion trajectories
- camera motion
- target object
- image sequences
- video signals
- independently moving objects
- motion patterns
- motion segmentation
- moving objects
- optical flow
- motion detection
- video shots
- motion capture
- video clips
- video streams
- background subtraction
- video analysis
- aerial video
- fuzzy logic
- traffic scenes
- fault diagnosis
- motion model
- image motion
- multimedia
- video surveillance
- human motion
- symbolic descriptions
- reference frame
- multi view
- moving object detection
- video objects
- visual features
- motion parameters
- prediction error
- high resolution
- spatio temporal
- interesting events
- aerial images
- d scene