A Modality Converting Approach for Image Annotation to Overcome the Inconsistent Labels in Training Data.
Tokinori SuzukiDaisuke IkedaPublished in: ICPR Workshops (2) (2020)
Keyphrases
- image annotation
- training data
- multi modal
- multi label
- automatic image annotation
- class labels
- multi label classification
- training set
- multi instance learning
- image search
- image annotation and retrieval
- visual features
- image content
- image understanding
- multiple instance learning
- web images
- training examples
- data sets
- image retrieval
- visual and textual information
- classification accuracy
- decision trees
- supervised learning
- training samples
- low level features
- labeled data
- learning algorithm
- semantic concepts
- unlabeled data
- web image annotation
- image collections
- graph cuts
- support vector machine
- low level
- label information
- prior knowledge
- annotated images
- semi supervised learning
- text categorization
- text classification
- image classification
- text mining
- support vector