The Cells Out of Sample (COOS) dataset and benchmarks for measuring out-of-sample generalization of image classifiers.
Alex X. LuAmy X. LuWiebke SchormannMarzyeh GhassemiDavid W. AndrewsAlan M. MosesPublished in: NeurIPS (2019)
Keyphrases
- input image
- image content
- image dataset
- single image
- multiscale
- image data
- image representation
- image segmentation
- image features
- image collections
- image set
- spatial information
- image classification
- decision trees
- image regions
- feature selection
- image retrieval
- microscopy images
- feature points
- low level
- high resolution
- image analysis
- pixel values
- million images
- naive bayes
- fold cross validation
- extracted features
- image pixels
- image matching
- segmentation method
- benchmark datasets
- segmentation algorithm