Conceptual 12M: Pushing Web-Scale Image-Text Pre-Training To Recognize Long-Tail Visual Concepts.
Soravit ChangpinyoPiyush SharmaNan DingRadu SoricutPublished in: CoRR (2021)
Keyphrases
- visual concepts
- web scale
- web images
- image content
- image collections
- image search
- long tail
- image features
- input image
- image annotation
- image classification
- multiscale
- semantic gap
- visual content
- visual features
- high resolution
- image retrieval
- image representation
- image data
- video content
- low level
- keywords
- semantic concepts
- positive examples
- image segmentation
- image matching
- multi label
- natural language processing