3DiVi Releases Face SDK 3.27 with Deepfake Detector, Turbo Inference with NVIDIA TensorRT and Intel OpenVINO, and Python Without GIL
3DiVi announces the release of Face SDK 3.27, a major update delivering a combination of security and system performance improvements. This latest version integrates a built-in deepfake detection module, alongside support for NVIDIA TensorRT, Intel OpenVINO acceleration, and no-GIL Python API — enabling fast, scalable, and attack-resistant face recognition solutions from day one.
Covina, CA, August 06, 2025 --(PR.com)-- Key Updates Include:
Deepfake Detection: Essential for 2025 Biometric Security
Deepfake attacks have become a real and growing threat worldwide, especially for remote identity verification scenarios. To counter this, 3DiVi introduces the new DEEPFAKE_ESTIMATOR processing block — a vital tool for KYC, digital onboarding, and cybersecurity teams aiming to stay ahead of increasingly sophisticated spoofing attempts.
TensorRT Support: Maximize NVIDIA GPU Performance
All Processing Block API modules now support inference acceleration with TensorRT on NVIDIA GPUs. This upgrade delivers peak performance with zero extra tuning or third-party tools — turbocharging your entire face recognition pipeline instantly.
OpenVINO Acceleration: Up to 50% Faster on Intel CPUs
With full integration of the OpenVINO framework in all main processing blocks, face recognition pipelines running on Intel CPUs with AVX512/VNNI instructions can experience up to a 50% speed increase — no GPU required.
Liveness Estimator Upgrade: 30% Better Spoof Detection Accuracy
A new version of the LIVENESS_ESTIMATOR Processing Block (2d_ensemble variant), delivers a 30% improvement in spoof attack detection accuracy compared to the previous version. This makes face anti-spoofing even more robust in real-world conditions — whether you're dealing with printed photos, replay attacks, or masks.
Python API without GIL: True Parallelism for Developers
Python developers can now use the no-GIL mode in the Processing Block API, enabling efficient parallelization of face recognition pipelines for higher system responsiveness and throughput.
End of Support for CUDA 10.2
To keep up with modern computing standards, 3DiVi is dropping support for CUDA 10.2 (released nearly six years ago). The SDK now runs on CUDA 11.x and 12.x, ensuring compatibility with the latest hardware and software environments.
Face SDK 3.27 is available now. Upgrade to benefit from faster, more scalable, and more secure face recognition technology for your projects.
Full documentation here: https://docs.3divi.ai/face_sdk/overview
More details: https://3divi.ai/products/software/face-sdk
Deepfake Detection: Essential for 2025 Biometric Security
Deepfake attacks have become a real and growing threat worldwide, especially for remote identity verification scenarios. To counter this, 3DiVi introduces the new DEEPFAKE_ESTIMATOR processing block — a vital tool for KYC, digital onboarding, and cybersecurity teams aiming to stay ahead of increasingly sophisticated spoofing attempts.
TensorRT Support: Maximize NVIDIA GPU Performance
All Processing Block API modules now support inference acceleration with TensorRT on NVIDIA GPUs. This upgrade delivers peak performance with zero extra tuning or third-party tools — turbocharging your entire face recognition pipeline instantly.
OpenVINO Acceleration: Up to 50% Faster on Intel CPUs
With full integration of the OpenVINO framework in all main processing blocks, face recognition pipelines running on Intel CPUs with AVX512/VNNI instructions can experience up to a 50% speed increase — no GPU required.
Liveness Estimator Upgrade: 30% Better Spoof Detection Accuracy
A new version of the LIVENESS_ESTIMATOR Processing Block (2d_ensemble variant), delivers a 30% improvement in spoof attack detection accuracy compared to the previous version. This makes face anti-spoofing even more robust in real-world conditions — whether you're dealing with printed photos, replay attacks, or masks.
Python API without GIL: True Parallelism for Developers
Python developers can now use the no-GIL mode in the Processing Block API, enabling efficient parallelization of face recognition pipelines for higher system responsiveness and throughput.
End of Support for CUDA 10.2
To keep up with modern computing standards, 3DiVi is dropping support for CUDA 10.2 (released nearly six years ago). The SDK now runs on CUDA 11.x and 12.x, ensuring compatibility with the latest hardware and software environments.
Face SDK 3.27 is available now. Upgrade to benefit from faster, more scalable, and more secure face recognition technology for your projects.
Full documentation here: https://docs.3divi.ai/face_sdk/overview
More details: https://3divi.ai/products/software/face-sdk
Contact
3DiVi Inc.
David Mitchell
+7 (351) 211-5099
3divi.ai
David Mitchell
+7 (351) 211-5099
3divi.ai
Contact
Categories