LATex is a novel framework that improves person identification across aerial and ground camera views by leveraging semantic attribute information without expensive model training.

• Integrates text-based attribute knowledge to enhance visual recognition capabilities
• Employs a more cost-effective training approach compared to full fine-tuning of large models
• Achieves superior identification accuracy for security and surveillance applications
• Creates more robust view-invariant features to match individuals across drastically different perspectives

Security Impact: This technology significantly enhances surveillance systems by enabling reliable tracking of individuals across drone and ground-level camera networks, critical for public safety and secure facilities monitoring.

LATex: Leveraging Attribute-based Text Knowledge for Aerial-Ground Person Re-Identification