Exploiting saliency for object segmentation from image level labels.
Seong Joon OhRodrigo BenensonAnna KhorevaZeynep AkataMario FritzBernt SchielePublished in: CoRR (2017)
Keyphrases
- object segmentation
- segmentation method
- image segmentation
- pixel level
- segmentation algorithm
- foreground objects
- graph cuts
- image features
- image data
- multiscale
- image analysis
- input image
- superpixels
- image content
- object detection
- test images
- image retrieval
- low level
- high resolution
- ground truth
- active contours
- image representation
- image regions
- computer vision
- image classification
- shape prior
- training set
- statistical model
- object categories
- saliency map
- pairwise
- image database
- small number
- figure ground