Weakly Supervised Few-shot Object Segmentation using Co-Attention with Visual and Semantic Embeddings.
Mennatullah SiamNaren DoraiswamyBoris N. OreshkinHengshuai YaoMartin JägersandPublished in: IJCAI (2020)
Keyphrases
- object segmentation
- weakly supervised
- superpixels
- visual features
- saliency map
- object detection
- object class
- graph cuts
- image segmentation
- low level features
- segmentation algorithm
- segmentation method
- visual information
- high level
- shape prior
- bounding box
- object categories
- visual attention
- object classes
- multiscale
- natural images
- semi supervised
- video sequences
- object recognition
- semantic information
- image retrieval
- keywords
- learning algorithm
- energy function
- video data
- feature vectors
- low level
- input image