Weakly Supervised Few-shot Object Segmentation using Co-Attention with Visual and Semantic Inputs.
Mennatullah SiamNaren DoraiswamyBoris N. OreshkinHengshuai YaoMartin JägersandPublished in: CoRR (2020)
Keyphrases
- object segmentation
- weakly supervised
- superpixels
- visual features
- saliency map
- object detection
- object class
- low level features
- image segmentation
- graph cuts
- shape prior
- segmentation algorithm
- segmentation method
- semantic information
- visual information
- low level
- visual attention
- high level
- higher level
- object categories
- bounding box
- object classes
- video sequences
- keywords
- key frames
- image database
- color histogram
- multiscale
- semi supervised
- computer vision
- viewpoint
- energy function
- level set
- image classification
- multi class
- input image