FaceVerifier is a real-time face recognition library built on OpenCV and DeepCore, (Deepixel’s proprietary library). It leverages TensorFlow Lite models for fast, anti-spoofing-enabled face recognition.
The system processes the entire pipeline from face detection to final verification in real time and provides:
- Real-time face detection and high-speed alignment
- Robust anti-spoofing functionality to block presentation attacks (such as photos and display screens)
- CPU-only optimized design for low-latency, standalone operations
The illustration below demonstrates how enrolled faces are authenticated.
- Real-time face detection and alignment
- Built-in Anti-spoofing (Presentation Attack Detection)
- High-speed face verification
- CPU-only operation (Optimized for edge/desktop environments)
(To be updated)
FaceVerifieris optimized for real-time performance on CPU. Typical inference speeds:
| Metric | Performance | Condition / Remark |
|---|---|---|
| Face Recognition Rate | 99.1% | - |
| Anti-Spoofing | 97.8% | TPR @ FPR |
Typical inference speed:
| Environment | Resolution | Latency |
|---|---|---|
| Apple M1 | 31.7 ms | |
| Galaxy S9 | 51.13 ms |
(To be updated)
(To be updated)
This library is proprietary and requires a paid license. You may not use, distribute, or modify it without a valid license.
- Contact us via email: support@deepixel.xyz
The face image used in this README/demo was generated by Nano Banana. This image is synthetic and does not depict a real person.