Multi-Target Tracking with Transferable Convolutional Neural Networks

TL;DR

This paper introduces a transferable CNN architecture for multi-target tracking that models the problem as image-to-image prediction, enabling large-scale tracking with improved performance and theoretical generalization bounds.

Contribution

It presents a novel transferable CNN framework for MTT, allowing effective scaling from small to large areas without re-training, supported by new theoretical analysis.

Findings

Outperforms random finite set filters on 10 targets

Successfully transfers to 250 targets with 29% performance gain

Provides theoretical bounds on generalization error

Abstract

Multi-target tracking (MTT) is a classical signal processing task, where the goal is to estimate the states of an unknown number of moving targets from noisy sensor measurements. In this paper, we revisit MTT from a deep learning perspective and propose a convolutional neural network (CNN) architecture to tackle it. We represent the target states and sensor measurements as images and recast the problem as an image-to-image prediction task. Then we train a fully convolutional model at small tracking areas and transfer it to much larger areas with numerous targets and sensors. This transfer learning approach enables MTT at a large scale and is also theoretically supported by our novel analysis that bounds the generalization error. In practice, the proposed transferable CNN architecture outperforms random finite set filters on the MTT task with 10 targets and transfers without re-training to a larger MTT task with 250 targets with a 29% performance improvement.