Multiple-target tracking in human and machine vision

Area of Science:

• Cognitive Neuroscience
• Computer Vision
• Artificial Intelligence

Background:

• Human visual perception enables simultaneous tracking of multiple objects during daily activities.
• Existing computer vision algorithms for multiple-target tracking (MTT) are applied in surveillance, sports analysis, and human-computer interaction.
• Despite advancements in AI, human MTT capabilities remain largely unimitated in computational models.

Purpose of the Study:

• To review multiple-target tracking (MTT) research in neuroscience.
• To examine biologically inspired MTT methods in computer vision.
• To discuss the complementary relationship between human MTT and AI approaches.

Main Methods:

• Literature review of neuroscience studies on human multiple-target tracking.
• Survey of artificial intelligence algorithms for automated multiple-target tracking.
• Comparative analysis of human and computational MTT mechanisms.

Main Results:

• Neuroscience reveals complex behavioral and neural mechanisms underlying human MTT.
• Computer vision offers various algorithms for automated MTT with applications in diverse fields.
• A significant gap exists between human MTT proficiency and current AI imitation.

Conclusions:

• Neuroscience and computer vision offer complementary perspectives on multiple-target tracking.
• Integrating insights from human cognition could enhance artificial intelligence MTT algorithms.
• Further research can bridge the gap between biological and artificial MTT systems.