Figure-Ground Image Segmentation Helps Weakly-Supervised Learning of Objects.
Katerina FragkiadakiJianbo ShiPublished in: ECCV (6) (2010)
Keyphrases
- figure ground
- image segmentation
- weakly supervised
- object segmentation
- long range
- low level
- object appearance
- graph cuts
- natural images
- image processing
- pixel level
- deformable models
- multiscale
- moving objects
- active contours
- object classes
- computer vision
- mean shift
- spatial information
- multi class
- shape prior
- viewpoint
- object class
- training data
- high level