VL-BERT: Pre-training of Generic Visual-Linguistic Representations.
Weijie SuXizhou ZhuYue CaoBin LiLewei LuFuru WeiJifeng DaiPublished in: ICLR (2020)
Keyphrases
- training process
- training phase
- domain specific
- higher level
- visual perception
- semantic representations
- natural language processing
- visual features
- category specific
- visual information
- visual representations
- computer vision
- training samples
- neural network
- low level
- training set
- natural language
- training algorithm
- visual cues
- virtual laboratory
- training examples
- visual patterns
- mid level
- test set
- image classification
- high level
- machine learning