RepMet: Representative-based metric learning for classification and one-shot object detection.
Eli SchwartzLeonid KarlinskyJoseph ShtokSivan HararyMattias MarderSharathchandra PankantiRogério Schmidt FerisAbhishek KumarRaja GiryesAlexander M. BronsteinPublished in: CoRR (2018)
Keyphrases
- metric learning
- object detection
- distance metric learning
- nearest neighbor classification
- distance metric
- machine learning and pattern recognition
- feature space
- margin maximization
- classification accuracy
- image classification
- semi supervised
- pattern recognition
- multi task
- pairwise
- support vector
- mahalanobis metric
- distance function
- computer vision
- machine learning
- support vector machine
- data sets
- face detection
- multi class
- background knowledge
- object categories
- training set
- reinforcement learning
- object class
- kernel learning
- gender classification
- feature extraction
- feature selection