A Cross-Modal Approach to Cleansing Weakly Tagged Images.
Yuejie ZhangChunlei YangYi ShenYuli GaoJianping FanHangzai LuoPublished in: IEEE Multim. (2010)
Keyphrases
- cross modal
- tagged images
- visual similarity
- semantic similarity
- visual features
- visual information
- automatic image annotation
- multi modal
- web images
- keywords
- visual content
- similarity measure
- image annotation
- visual concepts
- low level features
- dissimilarity measure
- learning algorithm
- semantic concepts
- wordnet
- key frames
- visual data
- image content
- semantic gap
- image classification
- co occurrence
- image data
- web pages