Image to Perturbation: An Image Transformation Network for Generating Visually Protected Images for Privacy-Preserving Deep Neural Networks.
Hiroki ItoYuma KinoshitaMaungMaung AprilPyoneHitoshi KiyaPublished in: IEEE Access (2021)
Keyphrases
- image transformations
- privacy preserving
- input image
- image features
- image data
- image collections
- image analysis
- image matching
- test images
- image retrieval
- image registration
- image pixels
- image classification
- data perturbation
- image regions
- keypoints
- pixel intensities
- rotation invariant
- sensitive data
- image content
- privacy preserving data mining
- image descriptors
- web images
- object recognition
- privacy protection
- pixel values
- dictionary learning
- pattern recognition
- image structure
- natural images
- feature points
- image representation
- differential privacy
- image processing
- image database
- salient regions
- data privacy
- visual features
- cbir systems
- sensitive information
- relevance feedback
- matching algorithm
- sift descriptors
- spatial information
- similarity measure
- scale space