Decoupled Box Proposal and Featurization with Ultrafine-Grained Semantic Labels Improve Image Captioning and Visual Question Answering.
Soravit ChangpinyoBo PangPiyush SharmaRadu SoricutPublished in: CoRR (2019)
Keyphrases
- question answering
- semantic labels
- image content
- web images
- image data
- low level
- information extraction
- image features
- information retrieval
- natural language
- image representation
- web pages
- news video
- visual data
- image collections
- named entities
- natural language processing
- image retrieval
- multiscale
- image classification
- image regions
- co occurrence
- text mining
- probabilistic model
- feature vectors
- visual information
- low level features
- relational databases
- visual content
- video sequences
- image sequences