Improvement of Gastroscopy Classification Performance Through Image Augmentation Using a Gradient-Weighted Class Activation Map.
Hyun-Sik HamHan-Sung LeeJung-Woo ChaeHyun Chin ChoHyun-chong ChoPublished in: IEEE Access (2022)
Keyphrases
- class specific
- image classification
- image data
- input image
- edge detection
- image features
- image analysis
- class labels
- image content
- gradient information
- image retrieval
- single image
- gradient direction
- image representation
- support vector machine
- high resolution
- preprocessing stage
- gradient field
- gradient orientation
- multi class classification
- intra class
- image regions
- test images
- gradient magnitude
- prior probabilities
- gradient domain
- feature points
- text classification
- classification accuracy
- pattern recognition
- support vector
- multiscale
- feature extraction
- image segmentation
- decision trees
- feature selection
- thematic mapper
- maximum a posterior
- maximum a posteriori probability
- image editing
- order statistics
- zero crossing
- hyperspectral images
- edge map
- vector field
- segmentation method
- image quality
- feature set
- denoising
- feature vectors
- training set
- image processing