Building embedded facial recognition is notoriously difficult. Between curating massive datasets, mitigating bias, and optimizing for constrained hardware, most projects stall in the laboratory phase and never reach production.
In this webinar, we will show you how to reduce the headache with a proven path to deployment. In collaboration with embedUR and STMicroelectronics, we will show you how to implement a complete, production-grade facial recognition pipeline entirely on a microcontroller with no cloud connectivity, no external compute, and no compromises.
During this one-hour session, we will explore ST’s new STM32N6 microcontroller, featuring dedicated AI acceleration and a vision-optimized memory architecture, paired with embedUR’s NovaEyeD AI stack. Attendees will be able to watch a live demonstration of the integration, featuring on-device face registration, real-time recognition, and physical boundary testing to demonstrate system robustness against varied angles and movement.
What you’ll learn:
- How ST’s new MCU class enables embedded teams to integrate advanced AI vision capabilities without utilizing external processors.
- How to execute a complete pipeline (detection, extraction, matching, and enrollment) locally to guarantee GDPR compliance and eliminate cloud infrastructure costs.
- Why edge AI breaks in the field, and how NovaEyeD solves those challenges with deterministic performance and known memory footprints.
- How the high efficiency of the NovaEyeD stack preserves MCU memory resources for parallel applications, such as access-control integration and presence detection.
Free