Visual Explanations of Image-Text Representations via Multi-Modal Information Bottleneck Attribution.
Ying WangTim G. J. RudnerAndrew Gordon WilsonPublished in: CoRR (2023)
Keyphrases
- multi modal
- auto annotation
- multiple modalities
- video search
- web images
- image annotation
- single modality
- cross modal
- input image
- automatic image annotation
- uni modal
- image analysis
- information bottleneck
- image content
- segmentation method
- information theoretic
- visual cues
- image features
- multi modality
- similarity measure
- image representation
- semantic concepts
- visual similarity
- image collections
- image classification
- low level
- image retrieval
- visual concepts
- image segmentation
- visual information