VL-BERT: Pre-training of Generic Visual-Linguistic Representations.
Weijie SuXizhou ZhuYue CaoBin LiLewei LuFuru WeiJifeng DaiPublished in: CoRR (2019)
Keyphrases
- visual information
- mid level
- semantic representations
- visual representations
- natural language
- training algorithm
- test set
- training examples
- high level
- training process
- serious games
- visual features
- higher level
- domain specific
- natural language processing
- data sets
- training samples
- visual perception
- information extraction
- linguistic knowledge
- low level
- category specific