Weakly-Supervised Visual-Textual Grounding with Semantic Prior Refinement.
Davide RigoniLuca ParolariLuciano SerafiniAlessandro SperdutiLamberto BallanPublished in: CoRR (2023)
Keyphrases
- weakly supervised
- semantic attributes
- relation extraction
- natural language
- superpixels
- topic models
- object class
- semantic features
- visual features
- high level
- visual information
- named entities
- information extraction
- multiscale
- viewpoint
- semantic similarity
- low level features
- semi supervised
- keywords
- domain specific
- semantic information
- learning algorithm
- object detection
- low level
- prior knowledge
- object recognition
- metadata