Making Higher Order MOT Scalable: An Efficient Approximate Solver for Lifted Disjoint Paths

TL;DR

This paper introduces an efficient approximate message passing solver for the NP-hard lifted disjoint paths problem, significantly improving scalability for large, crowded multiple object tracking sequences without compromising solution quality.

Contribution

It presents a novel approximate solver for LDP that enables scalable MOT in large, complex datasets, outperforming existing methods in efficiency and applicability.

Findings

Successfully scales to large MOT datasets like MOT20

Achieves comparable or better performance than state-of-the-art methods

Enables solving previously intractable large-scale MOT problems

Abstract

We present an efficient approximate message passing solver for the lifted disjoint paths problem (LDP), a natural but NP-hard model for multiple object tracking (MOT). Our tracker scales to very large instances that come from long and crowded MOT sequences. Our approximate solver enables us to process the MOT15/16/17 benchmarks without sacrificing solution quality and allows for solving MOT20, which has been out of reach up to now for LDP solvers due to its size and complexity. On all these four standard MOT benchmarks we achieve performance comparable or better than current state-of-the-art methods including a tracker based on an optimal LDP solver.