Improving task-specific representation via 1M unlabelled images without any extra knowledge.
Aayush BansalPublished in: CoRR (2020)
Keyphrases
- image data
- image database
- ground truth
- input image
- image classification
- image analysis
- edge detection
- object recognition
- image retrieval
- test images
- image representation
- feature representation
- domain knowledge
- representation language
- three dimensional
- knowledge base
- rigid body
- image registration
- acquisition process
- multiple images
- knowledge acquisition
- semi supervised learning
- image collections
- formal representation
- computer vision
- image annotation
- region of interest
- perceptual information
- image matching
- spatial information
- binary images
- medical images
- knowledge representation
- prior knowledge
- expert systems
- face recognition
- image sequences
- image segmentation