Weakly-Supervised Visual-Textual Grounding with Semantic Prior Refinement.
Davide RigoniLuca ParolariLuciano SerafiniAlessandro SperdutiLamberto BallanPublished in: BMVC (2023)
Keyphrases
- weakly supervised
- semantic attributes
- object class
- natural language
- relation extraction
- high level
- semi supervised
- topic models
- superpixels
- visual information
- object detection
- named entities
- semantic information
- metadata
- semantic features
- low level features
- domain specific
- natural language processing
- prior knowledge
- question answering
- co occurrence
- multi class
- information extraction
- object detectors
- low level
- information retrieval