A Unified Model for Tracking and Image-Video Detection Has More Power.
Peirong LiuRui WangPengchuan ZhangOmid PoursaeedYipin ZhouXuefei CaoSreya Dutta RoyAshish ShahSer-Nam LimPublished in: CoRR (2022)
Keyphrases
- unified model
- image frames
- video images
- face detection and tracking
- temporal continuity
- input image
- video scene
- image classification
- image data
- image content
- partial occlusion
- input video
- real time
- object detection and tracking
- single image
- segmentation method
- multiscale
- image segmentation
- person detection
- cluttered background
- object motion
- successive frames
- event detection
- complex background
- hough transform
- image retrieval
- accurate localization
- single frame
- surveillance videos
- target object
- foreground background separation
- video data
- multimedia
- image sequences
- object detection
- lighting conditions
- video sequences
- mean shift
- foreground background segmentation
- video surveillance
- edge detection
- video streams
- low level
- key frames
- human detection
- high resolution
- camera movement
- image features
- feature points
- face detection
- image regions
- motion model
- video frames
- visual concepts