Hafeez Anwar, Martin Kampel, and Rehanullah Khan
Attributes, Object Recognition, Image Retrieval
The recent use of attributes for various problems like object classification, object description and image retrieval produced promising results. The data sets are developed in which images are annotated with the attributes of objects instead of the category names. Attributes performed better than low-level features for cross-category generalization on several data sets. These data sets opened the avenues for potential research in computer vision and image under-standing based on attributes. Therefore it is quite necessary to get an overview of the attributes, the way they work and data sets annotated with attributes. In this work we provide an overview of various solutions based on attributes, the available data sets that are annotated with the attributes and future directions for some potential problems where at-tributes may perform better than low-level features.
Important Links:
Go Back
