CoCa: Contrastive Captioners are Image-Text Foundation Models.
Jiahui YuZirui WangVijay VasudevanLegg YeungMojtaba SeyedhosseiniYonghui WuPublished in: CoRR (2022)
Keyphrases
- image data
- single image
- input image
- random fields
- image features
- multiscale
- bayesian framework
- image pixels
- image analysis
- image content
- textual and visual information
- information retrieval
- image collections
- template matching
- image classification
- low level
- semantic information
- visual effects
- image statistics
- statistical model
- spatial information
- textual information
- image retrieval
- probabilistic model
- object recognition
- image representation
- bounding box
- image database
- lighting conditions
- edge detection
- image regions
- keypoints
- segmentation method