MONCE Tracking Metrics: a comprehensive quantitative performance evaluation methodology for object tracking

TL;DR

This paper introduces MONCE tracking metrics, a comprehensive set of quantitative tools designed to evaluate and diagnose the performance of complex multi-object tracking models, especially in challenging scenarios like non-contiguous and long-term tracking.

Contribution

The work extends existing benchmarks by proposing MONCE metrics that measure various aspects of multi-object tracking performance and provide diagnostic insights for model improvement.

Findings

MONCE metrics enable detailed performance evaluation.

Metrics include Expected Average Overlap and Re-Identification.

Provides diagnostic insights for tracking model development.

Abstract

Evaluating tracking model performance is a complicated task, particularly for non-contiguous, multi-object trackers that are crucial in defense applications. While there are various excellent tracking benchmarks available, this work expands them to quantify the performance of long-term, non-contiguous, multi-object and detection model assisted trackers. We propose a suite of MONCE (Multi-Object Non-Contiguous Entities) image tracking metrics that provide both objective tracking model performance benchmarks as well as diagnostic insight for driving tracking model development in the form of Expected Average Overlap, Short/Long Term Re-Identification, Tracking Recall, Tracking Precision, Longevity, Localization and Absence Prediction.