Image classifier learning from noisy labels via generalized graph smoothness priors.
Yu MaoGene CheungChia-Wen LinYusheng JiPublished in: IVMSP (2016)
Keyphrases
- unlabeled images
- training data
- learning process
- learning phase
- learning algorithm
- label noise
- multiscale
- image content
- single image
- high resolution
- image data
- image features
- input image
- supervised learning
- image retrieval
- auto annotation
- prior information
- image classification
- edge detection
- image segmentation
- segmentation method
- weakly supervised
- image representation
- labeled data
- image regions
- support vector machine
- active learning
- prior model
- labeled images
- learning stage
- prior knowledge
- bayes rule
- decision trees