Abstract
We developed a minimum-cost circulation framework for solving the global data association problem, which plays a key role in the tracking-by-detection paradigm of multi-object tracking (MOT). The global data association problem was extensively studied under the minimum-cost flow framework, which is theoretically attractive as being flexible and globally solvable. However, the high computational burden has been a long-standing obstacle to its wide adoption in practice. While enjoying the same theoretical advantages and maintaining the same optimal solution as the minimum-cost flow framework, our new framework has a better theoretical complexity bound and leads to orders of practical efficiency improvement. This new framework is motivated by the observation that minimum-cost flow only partially models the data association problem and it must be accompanied by an additional and time-consuming searching scheme to determine the optimal object number. By employing a minimum-cost circulation framework, we eliminate the searching step and naturally integrate the number of objects into the optimization problem. By exploring the special property of the associated graph, that is, an overwhelming majority of the vertices are with unit capacity, we designed an implementation of the framework and proved it has the best theoretical computational complexity so far for the global data association problem. We evaluated our method with 40 experiments on five MOT benchmark datasets. Our method was always the most efficient in every single experiment and averagely 53 to 1,192 times faster than the three state-of-the-art methods. When our method served as a sub-module for global data association methods utilizing higher-order constraints, similar running time improvement was attained. We further illustrated through several case studies how the improved computational efficiency enables more sophisticated tracking models and yields better tracking accuracy. We made the source code publicly available on GitHub with both Python and MATLAB interfaces.
Highlights
M ULTI-OBJECT tracking (MOT) is a fundamental task in machine intelligence with a variety of applications such as traffic surveillance, autonomous driving, particle tracking, and cell lineage analysis [1], [2]
(1) We proposed a new minimum-cost circulation-based framework for solving the maximum a posteriori (MAP) problem in MOT. (2) We developed a minimum-cost circulation algorithm and proved that it achieves the best ever theoretical bound for solving the MAP inference problem in MOT. (3) We implemented our proposed algorithm, which was shown to be empirically much more efficient than existing widely used methods in all of the 40 benchmark tests, often by several orders of improvement
It is worthy to mention that all the minimum-cost flow problems we encountered in MOT applications can be replaced by our minimum-cost circulation formulation, which means that all the existing minimum-cost flow-based identity inference models in MOT can be accelerated with the proposed formulation
Summary
M ULTI-OBJECT tracking (MOT) is a fundamental task in machine intelligence with a variety of applications such as traffic surveillance, autonomous driving, particle tracking, and cell lineage analysis [1], [2]. In the past ten years, one celebrated progress on the identity inference is the formulation of the MAP problems over multi frames as a minimum-cost flow problem [10], [12], [13], [14], [15], [16]. In real applications, minimum-cost flow-based identity inference models [12], [13], [14] are limited to datasets with a small number of detections or frames due to computational cost. Compared with the wide application of advanced affinity models, the majority of recent works on MOT still rely on simple and greedy data association strategies.
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: IEEE transactions on pattern analysis and machine intelligence
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
