Document Detail

Learning AND-OR Templates for Object Recognition and Detection.
MedLine Citation:
PMID:  23868779     Owner:  NLM     Status:  In-Data-Review    
This paper presents a framework for unsupervised learning of a hierarchical reconfigurable image template--the AND-OR Template (AOT) for visual objects. The AOT includes: 1) hierarchical composition as "AND" nodes, 2) deformation and articulation of parts as geometric "OR" nodes, and 3) multiple ways of composition as structural "OR" nodes. The terminal nodes are hybrid image templates (HIT) [17] that are fully generative to the pixels. We show that both the structures and parameters of the AOT model can be learned in an unsupervised way from images using an information projection principle. The learning algorithm consists of two steps: 1) a recursive block pursuit procedure to learn the hierarchical dictionary of primitives, parts, and objects, and 2) a graph compression procedure to minimize model structure for better generalizability. We investigate the factors that influence how well the learning algorithm can identify the underlying AOT. And we propose a number of ways to evaluate the performance of the learned AOTs through both synthesized examples and real-world images. Our model advances the state of the art for object detection by improving the accuracy of template matching.
Zhangzhang Si; Song-Chun Zhu
Related Documents :
24314859 - Detection of pigment network in dermoscopy images using supervised machine learning and...
23151349 - Application of cathodoluminescence microscopy and spectroscopy in geosciences.
24571349 - Immunochromatographic diagnostic test analysis using google glass.
24808039 - On the impact of approximate computation in an analog destin architecture.
24009659 - Equilibrium-phase high spatial resolution contrast-enhanced mr angiography at 1.5t in p...
23955749 - Adaptive dictionary learning in sparse gradient domain for image recovery.
20841889 - A web service for enabling medical image retrieval integrated into a social medical ima...
22660079 - Fast model-based multispectral imaging using nonnegative principal component analysis.
9241729 - Three dimensional structure of human fibrinogen under aqueous conditions visualized by ...
Publication Detail:
Type:  Journal Article    
Journal Detail:
Title:  IEEE transactions on pattern analysis and machine intelligence     Volume:  35     ISSN:  1939-3539     ISO Abbreviation:  IEEE Trans Pattern Anal Mach Intell     Publication Date:  2013 Sep 
Date Detail:
Created Date:  2013-07-22     Completed Date:  -     Revised Date:  -    
Medline Journal Info:
Nlm Unique ID:  9885960     Medline TA:  IEEE Trans Pattern Anal Mach Intell     Country:  United States    
Other Details:
Languages:  eng     Pagination:  2189-205     Citation Subset:  IM    
University of California, Los Angeles, Los Angeles.
Export Citation:
APA/MLA Format     Download EndNote     Download BibTex
MeSH Terms

From MEDLINE®/PubMed®, a database of the U.S. National Library of Medicine

Previous Document:  KNN Matting.
Next Document:  Modeling Natural Images Using Gated MRFs.