3DiVi News

Meet the New 3DiVi Face SDK with 40% Higher Liveness Quality, NPU Inference, Detector Cascades and More

Face biometrics in time and attendance, fintech, and access control need to be quick and precise to eliminate long wait times and minimize fraud risks.

The first 3DiVi Face SDK release of 2025 delivers just that—impressive boost in speed and accuracy, making face recognition more secure and adaptable for solutions of any scale, from mobile and desktop apps to professional server setups.

What’s New in Face SDK 3.25.0?

50% Faster Biometric Template Extraction & 60% Lighter Models

Speed matters. That’s why we’ve introduced lightweight versions for all key face recognition methods. With little impact on accuracy, these optimized models deliver:
  • 40-50% faster inference on mobile CPUs

  • 20-25% faster inference on desktop CPUs

  • 60% reduction in model weights for template extraction
This means quicker authentication, lower hardware costs, and a better user adoption for your tech.

40% Boost in Liveness Detection Accuracy

Fraudsters are getting smarter, and so is our Liveness Detection algorithm. The latest update increases attack detection accuracy by 40%, significantly reducing false positives (APCER@BPCER=0.05) across all attack types, ensuring higher security of remote identification operations.

Optimize your Complex Scenarios by Combining Face Detectors

In cases where you need to detect faces in the same entrypoint on different domain data, such as selfies, ACS camera photos and ID card photos, ensuring high quality can be challenging.

By combining multiple face detectors into a cascade, you won't miss a single face!

Support for Accelerated Inference on Rockchip NPU

The Face Detector and Face Template Extractor modules now support inference on Rockchip NPU. This new inference capability significantly speeds up calculations, even on low-performance devices.

Filter Out Small Faces by BBox Size

Refine your detections by adjusting width and height, perfect for removing irrelevant faces in the background.

A Clear-Cut Face Database Management

We redesigned TemplateIndex - the module responsible for storing the template database and 1:N searches. With our new TemplateIndex module, you can:
  • Dynamically (without recreating TemplateIndex) add and remove templates from the database.
  • Name each template with an arbitrary alphanumeric identifier (previously you had to store this information separately).
  • Use an asynchronous version of TemplateIndex for high-demand applications, enabling simultaneous template addition/removal and 1:N searches.
This update streamlines deployment and reduces downtime for large-scale systems.

C# Face SDK API Updated to .NET 8

With .NET 6's LTS expiration, we’ve migrated our C# API to .NET 8, ensuring our SDK stays compatible with the latest frameworks.

Upgrade Today and Stay Ahead

With Face SDK 3.25.0, you get faster face recognition performance, higher security, and smarter data management—critical for fintech, access control, and attendance systems.

Don’t let outdated technology slow you down. Contact us today to get a free trial!
2025-02-25 13:00 News Face SDK/API