Unsupervised Multiple-Object Tracking with a Dynamical Variational Autoencoder

TL;DR

This paper introduces DVAE-UMOT, an unsupervised probabilistic model using a dynamical variational autoencoder for multi-object tracking, which models object dynamics and outperforms existing probabilistic methods.

Contribution

The paper presents a novel unsupervised multi-object tracking approach using a dynamical variational autoencoder that models object dynamics after training on synthetic data.

Findings

DVAE-UMOT competes with state-of-the-art probabilistic MOT models.

It surpasses existing models in tracking performance.

The method is unsupervised and leverages variational inference.

Abstract

In this paper, we present an unsupervised probabilistic model and associated estimation algorithm for multi-object tracking (MOT) based on a dynamical variational autoencoder (DVAE), called DVAE-UMOT. The DVAE is a latent-variable deep generative model that can be seen as an extension of the variational autoencoder for the modeling of temporal sequences. It is included in DVAE-UMOT to model the objects' dynamics, after being pre-trained on an unlabeled synthetic dataset of single-object trajectories. Then the distributions and parameters of DVAE-UMOT are estimated on each multi-object sequence to track using the principles of variational inference: Definition of an approximate posterior distribution of the latent variables and maximization of the corresponding evidence lower bound of the data likehood function. DVAE-UMOT is shown experimentally to compete well with and even surpass the performance of two state-of-the-art probabilistic MOT models. Code and data are publicly available.