Weakly-Supervised Learning of Visual Relations in Multimodal Pretraining.
Emanuele BugliarelloAida NematzadehLisa Anne HendricksPublished in: EMNLP (2023)
Keyphrases
- weakly supervised learning
- weakly supervised
- object detection
- multiple instance learning
- spatial relations
- visual information
- object class
- multi modal
- multi view learning
- low level
- visual features
- learning algorithm
- natural images
- semantic relations
- domain specific
- supervised learning
- object recognition
- natural language
- feature selection