Facial Recognition System Requirements
Camera Integration:

Supports the connection of no fewer than 48 face capture cameras to receive image streams.
Facial Feature Extraction and Comparison:

Extracts facial features from captured images and compares them against all facial features in the watchlist, generating similarity scores.
Performs comparisons between captured faces and faces stored in various face libraries, including:
Dynamic face library (for live captured faces)
Static face library (for previously captured faces)
Registered normal personnel library
Registered abnormal personnel library
Displays comparison results, face images, and associated information based on user permissions.
Alert System:

Analyzes similarity scores from facial comparisons.
Triggers alarm notifications when similarity scores exceed the configured threshold.
Manual Management:

Authorized users can manually add, edit, and tag individuals in the dynamic and static face libraries.
Supports encrypted import of static face libraries for both registered normal and abnormal personnel.
Face Image Requirements:

All enrolled faces must be unobstructed images.
The system can recognize and compare faces even with partial obstructions, such as masks or hats, covering up to 50% of the face, and output corresponding alerts.
Face Library Capacity:

The system can manage and compare against a facial database with no fewer than 300,000 individuals.
Detection and Accuracy:

Face detection rate: ≥ 99% under standard conditions.
False positive rate (for non-watchlist faces): ≤ 0.5%.
Missed detection rate (for watchlist faces): ≤ 2%.