3DiVi Inc. Announces the Release of Face SDK 3.25.0 for Server, Mobile and Embedded Solutions
3DiVi Inc. has released Face SDK 3.25.0, the first major update of 2025, with higher speed and accuracy rates for face recognition in fintech, time and attendance, and access control. Key improvements include 50% faster biometric template extraction, 40% better liveness detection accuracy, and support for detector cascades. The SDK also features accelerated inference on Rockchip NPU, small face filtering, redesigned database management for dynamic operations, and a C# API updated to .NET 8.
Walnut, CA, February 28, 2025 --(PR.com)-- 3DiVi announces the release of Face SDK 3.25.0, the first major update in 2025, with speed and accuracy improvements to face recognition technology. This new version is designed to meet the high demands of mobile, desktop, and server-based solutions, particularly in fintech, time and attendance, and access control, where precision and quick response times are critical for reducing fraud and improving user experience.
The Key Updates
50% Faster Biometric Template Extraction and 60% Lighter Models:
With optimized versions of all key face recognition methods, Face SDK 3.25.0 brings 40-50% faster inference on mobile CPUs, 20-25% faster inference on desktop CPUs for template extraction and 60% reduction in model weights used for template extraction.
40% Boost in Liveness Detection Accuracy:
Liveness Detection algorithm provides a 40% increase in attack detection accuracy, reducing false positives (APCER@BPCER=0.05) in remote identification scenarios.
Optimized Complex Scenarios with Detector Cascades:
For scenarios with face detection in the same entrypoint on different domain data, such as selfies, ACS camera photos and ID card photos, Face SDK 3.25.0 introduces the ability to combine multiple face detectors into a cascade, ensuring high-quality detection across diverse data sources.
Support for Accelerated Inference on Rockchip NPU:
With the addition of Rockchip NPU support, the Face Detector and Face Template Extractor modules now offer calculations on low-performance devices.
Small Faces Filtered Out by BBox Size:
A new feature to filter out irrelevant small faces in the background by adjusting the bounding box size.
Redesigned Face Database Management:
The redesigned TemplateIndex module now allows dynamic template addition and removal without downtime, plus support for asynchronous operations to handle high-demand applications with simultaneous template 1:N searches.
C# Face SDK API Updated to .NET 8:
In line with the expiration of .NET 6’s LTS, 3DiVi has updated its C# Face SDK API to .NET 8, ensuring compatibility with the latest frameworks.
About 3DiVi Inc.
3DiVi Inc., founded in 2011, is one of the leading international developers of AI and machine learning (ML) technologies for computer vision. The company's computer vision algorithms cover face recognition, body and skeletal tracking, spatial understanding and object recognition.
For more information: https://3divi.ai/
The Key Updates
50% Faster Biometric Template Extraction and 60% Lighter Models:
With optimized versions of all key face recognition methods, Face SDK 3.25.0 brings 40-50% faster inference on mobile CPUs, 20-25% faster inference on desktop CPUs for template extraction and 60% reduction in model weights used for template extraction.
40% Boost in Liveness Detection Accuracy:
Liveness Detection algorithm provides a 40% increase in attack detection accuracy, reducing false positives (APCER@BPCER=0.05) in remote identification scenarios.
Optimized Complex Scenarios with Detector Cascades:
For scenarios with face detection in the same entrypoint on different domain data, such as selfies, ACS camera photos and ID card photos, Face SDK 3.25.0 introduces the ability to combine multiple face detectors into a cascade, ensuring high-quality detection across diverse data sources.
Support for Accelerated Inference on Rockchip NPU:
With the addition of Rockchip NPU support, the Face Detector and Face Template Extractor modules now offer calculations on low-performance devices.
Small Faces Filtered Out by BBox Size:
A new feature to filter out irrelevant small faces in the background by adjusting the bounding box size.
Redesigned Face Database Management:
The redesigned TemplateIndex module now allows dynamic template addition and removal without downtime, plus support for asynchronous operations to handle high-demand applications with simultaneous template 1:N searches.
C# Face SDK API Updated to .NET 8:
In line with the expiration of .NET 6’s LTS, 3DiVi has updated its C# Face SDK API to .NET 8, ensuring compatibility with the latest frameworks.
About 3DiVi Inc.
3DiVi Inc., founded in 2011, is one of the leading international developers of AI and machine learning (ML) technologies for computer vision. The company's computer vision algorithms cover face recognition, body and skeletal tracking, spatial understanding and object recognition.
For more information: https://3divi.ai/
Contact
3DiVi Inc.
David Mitchell
+1 408 462 2760
3divi.ai
David Mitchell
+1 408 462 2760
3divi.ai
Contact
Categories