Summary
the 2014 International Symposium on Nonlinear Theory and its Applications
2014
Session Number:D2L-B
Session:
Number:D2L-B3
A 2-Fan Shape Model for Object Detection and Localization
Yuanqi Su, Yuehu Liu, Xiao Huang, Nanning Zheng,
pp.759-762
Publication Date:2014/9/14
Online ISSN:2188-5079
Summary:
For its simplicity and effectiveness, star model is popular in part-based object detection and localization. It finds the candidates for parts and then vote for object’s configurations. However, it su?ers from the loose geometric constraints since it neglects the connection among parts. In the paper, we reconsidered the connections and reduce the task of object detection and localization to a shape matching problem. We propose to use a global shape as a constraint to modify the star model. The resulting representation is then a 2-fan model that is subsequently optimized by searching for the candidates of parts and solving for the global constraint. Though the procedure is easily stuck at a local optimum, results show that is gives a comparable result with the state-of-art methods.