文章基本信息

标题：FAST OBJECT DETECTION WITH DEFORMABLE PART MODELS BASED ON HIERARCHICAL MODEL
本地全文：下载
作者：LIU MENG ; QINGXUAN JIA
期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2012
卷号：42
期号：1
页码：142-149
出版社：Journal of Theoretical and Applied
摘要：We propose a deformable parts model attached the coarse to fine detection methods to improve the detection speed. General when we meet the complex target structure, image changes, background interference problems, contour information of the target problems are often difficult to accurately describe. In order to solve the problem, we propose a hierarchical model: first we introduce the Conditional Random Fields (CRF) in the low-layer to fuse sorts of features which can obtain the target candidate region for high-layer that auxiliary to provide the appearance feature; in high-level we molding the contour features of the model and deformable parts model as the underlying detector, the model training and testing use of coarse to fine inference, so it will not only improve the detection speed but also improve the accuracy rate .The experiments on the PASCAL 2007 dataset show that the model flexibility, the universal recognition results are accurate and stable.
关键词：Deformable Parts Model; Coarse to Fine Detection; Conditional Random Fields