Human-interpretable and deep features for image privacy classification.
Darya BaranouskayaAndrea CavallaroPublished in: ICIP (2023)
Keyphrases
- image features
- extracting features
- extracted features
- image classification
- textural features
- feature representation
- feature values
- feature extraction
- class specific
- feature vectors
- low level
- feature space
- image data
- classification accuracy
- original images
- feature set
- image representation
- static images
- spectral features
- input image
- classification models
- test images
- svm classifier
- visual object recognition
- individual features
- gabor filtering
- image analysis
- feature histograms
- support vector machine svm
- invariant features
- multiscale
- keypoints
- spatial information
- image content
- classification method
- image segmentation
- classification rules
- segmentation method
- privacy preserving
- classification process
- human observers
- confidence scores
- highly discriminative
- image matching
- image retrieval
- texture analysis
- feature selection
- high resolution
- image pixels
- single image
- single feature
- image processing
- object recognition
- visual attributes
- image regions
- global features
- image set
- personal information
- feature points
- edge detection
- skin lesion
- feature descriptors
- training set
- multiple features
- similarity measure
- decision trees
- gray level co occurrence matrix