Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Video

Updated: Jul 11, 2026

Utilizing vmTracking to Improve the Accuracy of Multi-Animal Pose Estimation in Rodent Social Behavior Studies

Utilizing vmTracking to Improve the Accuracy of Multi-Animal Pose Estimation in Rodent Social Behavior Studies

Published on: November 7, 2025

BM3 E: discriminative density propagation for visual tracking.

Cristian Sminchisescu¹, Atul Kanaujia, Dimitris N Metaxas

¹Toyota Technological Institute-Chicago, University of Chicago, 1427 East 60th Street, Second Floor, Chicago, IL 60637, USA. crismin@nagoya.uchicago.edu

IEEE Transactions on Pattern Analysis and Machine Intelligence

|September 13, 2007

Summary

We introduce BM3 E, a Conditional Bayesian Mixture of Experts Markov Model, for consistent probabilistic estimates in discriminative visual tracking. This new model offers improved performance in complex state distribution prediction and human motion reconstruction.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

Toward Modality- and Sampling-Universal Learning Strategies for Accelerating Cardiovascular Imaging: Summary of the CMRxRecon2024 Challenge.

IEEE transactions on medical imaging·2025

Same author

Reconstructing Three-Dimensional Models of Interacting Humans.

IEEE transactions on pattern analysis and machine intelligence·2025

Same author

The state-of-the-art in cardiac MRI reconstruction: Results of the CMRxRecon challenge in MICCAI 2023.

Medical image analysis·2025

Same author

Classification of lung cancer subtypes on CT images with synthetic pathological priors.

Medical image analysis·2024

Same author

A Critical Analysis of the Limitation of Deep Learning based 3D Dental Mesh Segmentation Methods in Segmenting Partial Scans.

Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Annual International Conference·2023

Same author

Automatic Lung Cancer Subtypes Classification on CT Images with Self-generated Multi-modality Hybrid Features.

Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Annual International Conference·2023

Area of Science:

Computer Vision
Machine Learning
Probabilistic Modeling

Background:

Generative models using Kalman or particle filtering are common for visual tracking.
Existing methods often require inverting non-linear generative observation models at runtime.
There is a need for bottom-up approaches to complement existing top-down models.

Purpose of the Study:

To introduce BM3 E, a novel Conditional Bayesian Mixture of Experts Markov Model for discriminative visual tracking.
To develop a model capable of consistent probabilistic estimates and temporal inference.
To provide a bottom-up alternative to generative models in visual tracking.

Main Methods:

Utilizing a Conditional Bayesian Mixture of Experts Markov Model (BM3 E).

More Related Videos

Trajectory Data Analyses for Pedestrian Space-time Activity Study

Trajectory Data Analyses for Pedestrian Space-time Activity Study

Published on: February 25, 2013

A Protocol for Real-time 3D Single Particle Tracking

A Protocol for Real-time 3D Single Particle Tracking

Published on: January 3, 2018

Related Experiment Videos

Last Updated: Jul 11, 2026

Utilizing vmTracking to Improve the Accuracy of Multi-Animal Pose Estimation in Rodent Social Behavior Studies

Utilizing vmTracking to Improve the Accuracy of Multi-Animal Pose Estimation in Rodent Social Behavior Studies

Published on: November 7, 2025

Trajectory Data Analyses for Pedestrian Space-time Activity Study

Trajectory Data Analyses for Pedestrian Space-time Activity Study

Published on: February 25, 2013

A Protocol for Real-time 3D Single Particle Tracking

A Protocol for Real-time 3D Single Particle Tracking

Published on: January 3, 2018

Learning to predict complex state distributions directly from image observation descriptors (e.g., histograms, spatial grids).

Integrating descriptors into a conditional graphical model for temporal smoothness and uncertainty management.

Employing sparsity, mixture modeling, and non-linear dimensionality reduction for efficient computation.

Main Results:

Established density propagation rules for discriminative inference in continuous, temporal chain models.
Developed supervised and unsupervised algorithms for learning feedforward, multivalued contextual mappings.
Empirically validated the framework for 3D human motion reconstruction from monocular video.
Demonstrated significant performance gains over competing methods.

Conclusions:

BM3 E provides consistent probabilistic estimates for discriminative visual tracking.
The model effectively handles temporal and uncertain inference.
The framework shows superior performance in 3D human motion reconstruction, outperforming existing methods.