Cycle-Correspondence Loss: Learning Dense View-Invariant Visual Features from Unlabeled and Unordered RGB Images.
David B. AdrianAndras Gabor KupcsikMarkus SpiesHeiko NeumannPublished in: CoRR (2024)
Keyphrases
- visual features
- image classification
- image collections
- image retrieval
- visual appearance
- image annotation
- learning algorithm
- web images
- supervised learning
- visual data
- visual content
- image data
- visual information
- image search
- labeled images
- semantic gap
- input image
- image database
- visual similarity
- low level visual features
- human actions
- view invariant
- multi view
- active learning
- object recognition
- three dimensional
- action recognition
- image features
- bag of features
- keywords
- image set
- point correspondences
- image registration
- low level