Minimum Cost Multi-Way Data Association for Optimizing Multitarget Tracking of Interacting Objects.

Area of Science:

• Computer Science
• Engineering
• Data Science

Background:

• Existing multitarget tracking methods often struggle with complex object interactions such as merging or splitting.
• Tracking systems need to handle general degrees of inter-object interactions for real-world applications.

Purpose of the Study:

• To develop a general formulation for minimum cost data association problems.
• To address multitarget tracking challenges with general degrees of inter-object interactions.
• To enable tracking of complex phenomena like aggregating or splitting nanoparticles.

Main Methods:

• Formulated the data association problem as a binary integer programming problem.
• Developed a polynomial-time solution approach providing a good relaxation solution.
• Validated the method on simulated and real-world multitarget tracking scenarios.

Main Results:

• The proposed formulation effectively handles one-to-one, m-to-one, and one-to-n data associations.
• The polynomial-time solution yields integral results with improved duality gap compared to linear relaxation.
• The approach is applicable to moderate-sized multitarget tracking problems (hundreds of targets, tens of frames).

Conclusions:

• The general data association formulation provides a robust solution for complex multitarget tracking.
• The efficient solution method makes it practical for real-world applications with interacting targets.
• This work advances the capabilities of multitarget tracking systems in dynamic environments.