Offline AI Binocular 3D Vision Recognition Sensor (Face / Palm Vein / QR Code)

The AI Binocular Vision Recognition Sensor is a standalone, multi-modal identity recognition module designed for secure and flexible authentication applications. It features an onboard AI processor and uses dual RGB and infrared cameras to perform recognition tasks locally, without requiring processing power from the host controller.

This compact sensor combines face recognition, palm vein recognition, and QR code scanning in one module. It also supports 3D liveness detection to help prevent spoofing attempts using photos or videos. With offline AI processing and local storage for up to 1,000 face profiles and 1,000 palm vein templates, it is ideal for access control, smart home systems, smart locks, payment terminals, and automated retail applications.

High-Security Biometrics with 3D Liveness Detection

For authentication systems, reliable security is essential. This binocular vision sensor uses both RGB and infrared cameras to capture depth and liveness information. Its 3D liveness detection algorithm helps distinguish a real person from a flat image or video, providing stronger protection against spoofing attempts.

With a face recognition False Acceptance Rate as low as 0.001%, the module is suitable for applications that require dependable identity verification, such as secure door access, smart locks, and payment systems.

3-in-1 Recognition Capability

The sensor supports three different recognition methods in one compact device:

Face recognition provides fast, contactless user identification.

Palm vein recognition identifies unique vein patterns beneath the skin, offering a highly secure biometric method that is difficult to copy.

QR code recognition allows convenient temporary access, payment verification, device pairing, or user identification.

This combination makes the sensor suitable for a wide range of authentication and access control projects.

Offline AI Processing

The module is equipped with a dedicated AI System-on-Chip featuring a 0.5 TOPS Neural Processing Unit. All recognition algorithms are processed directly on the module, reducing the workload on the host controller.

Because processing and data storage are handled locally, the sensor can operate without an internet connection. This improves response time, enhances data privacy, and allows easy integration with microcontrollers such as Arduino or other embedded systems.

Easy Integration with UART and USB
The sensor outputs recognition results through a simple UART serial interface, making it easy to integrate into new or existing systems. It also supports USB UVC video transmission, allowing it to function as a standard video device for monitoring, visual confirmation, or video intercom applications.

Features

3-in-1 recognition: face, palm vein, and QR code
Binocular 3D liveness detection to resist spoofing by photos or videos
Onboard offline AI processing with no host-computing requirement
Local storage for up to 1,000 faces and 1,000 palm vein templates
Operates in complete darkness and outdoor non-direct sunlight conditions
UART communication for simple embedded integration
USB UVC video output for video streaming applications
Applications
Smart home systems
Smart door locks
Secure access control systems
Self-service retail payment terminals
3D face and palm vein payment systems
Palm vein authentication systems
Video intercom and identity verification devices

Specifications

Hardware Parameters
SoC: Arm CPU @ 900MHz, NPU @ 0.5 TOPS, RISC-V @ 600MHz
DDR2: 64MB
Flash: 32MB
Cameras: Dual 1/5” CMOS, 2MP, dual MIPI interfaces
Lens: 83° diagonal FOV, optimal focus distance 60cm
LED: IR 850nm, RGB 650nm, 90° illumination
USB: UVC video transmission, MJPEG output
Communication: UART @ 115200 baud
Power Parameters
Supply Voltage: 5V to 12V
Operating Current: 320–330mA @ 8V
Standby Current: 120–130mA in auto-detection mode

Shutdown Current: 0µA in non-auto-detection mode

Other Parameters

Operating Temperature: -20°C to +60°C
Storage Temperature: -30°C to +70°C
Humidity: 10% to 93% RH, non-condensing
Audio I/O: Expandable speaker/microphone support
Boot Time: 900ms to 2.5s, depending on stored user count
User Capacity: 1,000 faces and 1,000 palm vein templates
Algorithms: Binocular liveness detection, deep learning face recognition, palm vein recognition, QR code recognition
Liveness FAR: ≤1%
Liveness FRR: ≤1%
Face Recognition Pass Rate: 98.85%
Face Recognition FAR: 0.001%
Recognition Angle: ±20° pitch/yaw, supports multi-angle enrollment
Detection Range: Face 30–120cm; palm vein / QR code 15cm
Dimensions: 57.8 × 20 × 10.12mm

Package Includes

1 × AI Binocular Vision Recognition Sensor
1 × 1.25mm 4-pin to Dupont female connector cable
1 × USB data cable