A review of attribute relationship exploration and exploitation for improved object recognition

Fengyi Song, Yi Chen


Recently, attribute-based representation has been extensively researched in the literature of computer vision due to its wide applications. While enjoying advantages of attributes such as clear semantic meaning and well generality in depicting different classes, the inherent relationship among attributes should not be neglected. However, it is a nontrivial task of exploring and exploiting attribute relationship in real application scenarios, although some recent work has gained rich benefits from doing this. The goal of this paper is to categorize and discuss typical methods addressing this problem in a comprehensive way. We focus on overall difficulties and challenges in exploring attribute relationship, and summarize typical relationship among attributes and also typical approaches for relation exploitation. In addition, we will present a detailed survey of prominent algorithms from the view of explainable, tractable and their pros. and also cons.


Visual attribute; Attribute relationship; Survey

Full Text:



  • There are currently no refbacks.

Creative Commons License
This work is licensed under a Creative Commons Attribution 3.0 License.

JES ©: World Science Publisher United States