MoFM introduces a groundbreaking foundation model for semantic understanding of human movements in both time and space, following principles that drove LLM advancements.

• Trained on MotionBook, a comprehensive human motion dictionary
• Designed to understand and interpret complex movement patterns
• Enables generalization across diverse motion-related tasks
• Demonstrates scalability similar to successful language models

For security applications, MoFM offers significant potential in surveillance systems, anomaly detection, and behavior monitoring by accurately modeling unusual movement patterns that may indicate security threats.

MoFM: A Large-Scale Human Motion Foundation Model