Case Study: a global smart-home security provider improves facial recognition training with Defined.ai

Global Smart-home Security Provider delivers facial recognition data in 1 day with Defined.ai

A global smart-home security provider needed high-quality, diverse training data to improve the reliability of its facial recognition system under real-world conditions, such as low-light and partial visibility. They faced the challenge of collecting this data while integrating seamlessly with their existing devices and participant workflow. Defined.ai was engaged to provide a solution using consented video data collection.

Defined.ai implemented a protocol-driven capture using the customer's own equipment, recording 119 participants across different lighting conditions and scripted actions. The solution was delivered rapidly, with usable data provided on the very first day. The final, AI-ready dataset of surveillance-style video clips enabled the customer to immediately begin training their models for improved identification robustness and security-event understanding.