BURST: A Benchmark for Unifying Object Recognition, Segmentation and Tracking in Video.
Ali AtharJonathon LuitenPaul VoigtlaenderTarasha KhuranaAchal DaveBastian LeibeDeva RamananPublished in: CoRR (2022)
Keyphrases
- object segmentation and tracking
- object recognition
- text detection
- objects in video sequences
- foreground background segmentation
- video segmentation
- video scene
- real time
- multiple frames
- motion cues
- video images
- video sequences
- single frame
- motion segmentation
- video tracking
- berkeley segmentation dataset
- object detection and tracking
- video surveillance
- segmentation algorithm
- image segmentation
- object tracking
- face detection and tracking
- segmentation method
- object matching
- video analysis
- moving camera
- video data
- video dataset
- particle filter
- active contours
- automatic initialization
- level set
- temporal segmentation
- object recognition and tracking
- person detection
- figure ground segmentation
- computer vision
- video frames
- multimedia
- video objects
- video content
- shape prior
- bounding box
- surveillance videos
- dynamic scenes
- multiscale
- image features
- layered representation
- d objects
- motion estimation
- medical images
- natural images
- mean shift
- video streams
- appearance model
- object categories
- video clips
- temporal continuity
- complex background