Instance Segmentation and Tracking with Cosine Embeddings and Recurrent Hourglass Networks.
Christian PayerDarko SternThomas NeffHorst BischofMartin UrschlerPublished in: CoRR (2018)
Keyphrases
- vector space
- motion segmentation
- level set
- automatic initialization
- image segmentation
- real time
- social networks
- segmentation algorithm
- segmentation method
- deformable templates
- particle filtering
- low grade gliomas
- object segmentation
- region growing
- object tracking
- multiscale
- medical images
- particle filter
- euclidean distance
- occlusion reasoning
- image analysis
- automatic tracking
- occlusion detection
- fully unsupervised
- mean shift
- fully automatic
- feed forward
- accurate localization
- motion model
- energy function
- appearance model
- complex background
- multiple objects
- shape prior
- manifold learning
- bounding box
- motion tracking
- segmentation accuracy
- edge detection
- kalman filter
- euclidean space
- dimensionality reduction
- network structure
- feature vectors