Symbolizing Visual Features for Pre-training with Unlabeled Images.
Yuichi KamataMoyuru YamadaKeizo KatoAkira NakagawaTakayuki OkataniPublished in: ACPR (2) (2021)
Keyphrases
- visual features
- labeled images
- visual information
- visual content
- image classification
- image search
- image retrieval
- image annotation
- keywords
- image collections
- low level
- low level features
- semantic gap
- web images
- semantic features
- key frames
- machine learning
- nearest neighbor
- automatic image annotation
- image set
- training examples
- generative model
- cbir systems
- multiple features
- high level
- feature selection
- search engine