Sigmoid Loss for Language Image Pre-Training.
Xiaohua ZhaiBasil MustafaAlexander KolesnikovLucas BeyerPublished in: CoRR (2023)
Keyphrases
- image data
- multiscale
- single image
- image content
- input image
- image features
- image collections
- image segmentation
- image analysis
- image classification
- segmentation algorithm
- template matching
- image pixels
- image retrieval
- training set
- natural language
- pixel values
- similarity measure
- spatial information
- hough transform
- image regions
- image representation
- test images
- training examples
- language learning
- edge detection
- low level
- contrast enhancement