Weakly-supervised segmentation by combining CNN feature maps and object saliency maps.
Wataru ShimodaKeiji YanaiPublished in: ICPR (2016)
Keyphrases
- saliency map
- feature maps
- object segmentation
- superpixels
- visual attention
- input image
- human visual system
- bounding box
- natural images
- object class
- visual features
- multiple objects
- video frames
- object detectors
- segmentation algorithm
- eye movements
- self organizing maps
- low level features
- object classes
- eye tracking
- visual information
- machine learning
- image classification
- feature vectors
- image regions
- segmentation method
- image data
- training data
- image segmentation