ASIST: Annotation-free synthetic instance segmentation and tracking by adversarial simulations.
Quan LiuIsabella M. GaetaMengyang ZhaoRuining DengAadarsh JhaBryan A. MillisAnita Mahadevan-JansenMatthew J. TyskaYuankai HuoPublished in: Comput. Biol. Medicine (2021)
Keyphrases
- motion segmentation
- segmentation algorithm
- level set
- automatic initialization
- image segmentation
- segmentation method
- multiple objects
- particle filter
- medical images
- deformable templates
- automatic tracking
- occlusion detection
- semantic annotation
- multi agent
- real time
- foreground background separation
- multiple cues
- visual tracking
- accurate localization
- metadata
- complex background
- moving target
- shape prior
- kalman filter
- motion model
- image regions
- graph cuts
- active learning
- image analysis
- partial occlusion
- particle filtering
- object segmentation
- region growing
- motion tracking
- segmentation accuracy
- cluttered background
- fully automatic
- mean shift