Document Detail


Learning scene context for multiple object tracking.
MedLine Citation:
PMID:  19423445     Owner:  NLM     Status:  PubMed-not-MEDLINE    
Abstract/OtherAbstract:
We propose a framework for multitarget tracking with feedback that accounts for scene contextual information. We demonstrate the framework on two types of context-dependent events, namely target births (i.e., objects entering the scene or reappearing after occlusion) and spatially persistent clutter. The spatial distributions of birth and clutter events are incrementally learned based on mixtures of Gaussians. The corresponding models are used by a probability hypothesis density (PHD) filter that spatially modulates its strength based on the learned contextual information. Experimental results on a large video surveillance dataset using a standard evaluation protocol show that the feedback improves the tracking accuracy from 9% to 14% by reducing the number of false detections and false trajectories. This performance improvement is achieved without increasing the computational complexity of the tracker.
Authors:
Emilio Maggio; Andrea Cavallaro
Related Documents :
20307305 - Apples and oranges: avoiding different priors in bayesian dna sequence analysis.
6851495 - Computer simulations of learning in neural systems.
19639915 - Engaging learners across generations: the progressive professional development model.
11674845 - Predictability, complexity, and learning.
25017755 - Population survey and conservation assessment of the globally threatened cheer pheasant...
19163645 - Modeling transcranial dc stimulation.
Publication Detail:
Type:  Journal Article; Research Support, Non-U.S. Gov't     Date:  2009-05-05
Journal Detail:
Title:  IEEE transactions on image processing : a publication of the IEEE Signal Processing Society     Volume:  18     ISSN:  1057-7149     ISO Abbreviation:  IEEE Trans Image Process     Publication Date:  2009 Aug 
Date Detail:
Created Date:  2009-07-14     Completed Date:  2009-09-22     Revised Date:  -    
Medline Journal Info:
Nlm Unique ID:  9886191     Medline TA:  IEEE Trans Image Process     Country:  United States    
Other Details:
Languages:  eng     Pagination:  1873-84     Citation Subset:  -    
Affiliation:
Multimedia and Vision Group, Queen Mary University of London, London E14NS, UK. emilio.maggio@vicon.com
Export Citation:
APA/MLA Format     Download EndNote     Download BibTex
MeSH Terms
Descriptor/Qualifier:

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


Previous Document:  Mixture Model- and Least Squares-Based Packet Video Error Concealment.
Next Document:  Efficient block-based frequency domain wavelet transform implementations.