ZEST: Zero-shot Learning from Text Descriptions using Textual Similarity and Visual Summarization.
Tzuf Paz-ArgamanReut TsarfatyGal ChechikYuval AtzmonPublished in: EMNLP (Findings) (2020)
Keyphrases
- textual descriptions
- keywords
- image search
- textual information
- free text
- video search
- textual data
- visual and textual features
- visual information
- semantic labels
- visual representations
- web images
- natural language descriptions
- text content
- topic segmentation
- automatic summarization
- text analytics
- similarity measure
- text summarization
- visual features
- high level
- sentence similarity
- human generated
- textual features
- plain text
- knowledge transfer
- multimedia
- visual content
- lexical chains
- semantic content
- manually constructed
- text representation
- automatically created
- information retrieval
- visual similarity
- textual case based reasoning
- low level
- natural language
- text documents
- extractive summarization
- distance measure
- text mining
- information extraction
- mutual reinforcement
- multi document summarization
- news video
- text retrieval
- metadata
- automatic text summarization
- visual representation
- web documents
- web pages