ZEST: Zero-shot Learning from Text Descriptions using Textual Similarity and Visual Summarization.
Tzuf Paz-ArgamanYuval AtzmonGal ChechikReut TsarfatyPublished in: CoRR (2020)
Keyphrases
- textual descriptions
- web images
- image search
- textual data
- keywords
- video search
- visual features
- free text
- visual representations
- natural language descriptions
- textual information
- textual features
- visual information
- extractive summarization
- text summarization
- automatic summarization
- semantic content
- text content
- semantic similarity
- human generated
- similarity measure
- high level
- visual similarity
- topic segmentation
- text analytics
- semantic labels
- textual case based reasoning
- sentence similarity
- visual content
- news video
- text mining
- information retrieval
- automatically created
- low level
- knowledge transfer
- plain text
- text retrieval
- automatic text summarization
- image classification
- semantic concepts
- textual and visual information
- metadata
- information extraction
- edit distance
- multimedia
- word pairs
- web image search
- distance measure
- web pages
- text representation
- image content