Performance Evaluation Metrics and Approaches for Target Tracking: A Survey.

Area of Science:

• Signal Processing
• Control Systems
• Machine Learning

Background:

• Performance evaluation (PE) is critical for designing and validating target-tracking algorithms.
• Algorithm optimality is often defined by minimizing estimation errors, such as mean-squared error.
• Existing PE metrics and approaches require comprehensive review for current tracking challenges.

Purpose of the Study:

• To review classic and novel performance evaluation (PE) metrics for target tracking.
• To explore comprehensive evaluation (CE) approaches beyond traditional metrics.
• To demonstrate the application of PE metrics and CE approaches in target tracking scenarios.

Main Methods:

• Classification of common target tracking metrics into correctness, timeliness, and accuracy.
• Review of emerging comprehensive evaluation (CE) methods including cloud barycenter, fuzzy CE, and grey clustering.
• Demonstration of PE metrics and CE approaches in representative target tracking scenarios.

Main Results:

• Target tracking metrics are categorized into correctness, timeliness, and accuracy.
• Novel CE approaches like cloud barycenter, fuzzy CE, and grey clustering are presented.
• Practical application of PE metrics and CE approaches is illustrated through scenarios.

Conclusions:

• A structured review of PE metrics and CE approaches is provided for target tracking.
• The study highlights the importance of comprehensive evaluation for algorithm validation.
• The findings offer guidance for selecting and applying appropriate PE metrics and CE methods in target tracking.