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

The Offline AI Binocular 3D Vision Recognition Sensor is a standalone biometric module featuring face recognition, palm vein identification, and QR code decoding. Powered by an onboard 0.5 TOPS NPU and dual RGB-infrared cameras, it handles all AI processing and data storage locally without loading your host microcontroller. It is ideal for students, makers, and developers building secure access control, smart door locks, and interactive automation projects.

3D liveness anti-spoofing detection

The binocular RGB and infrared camera system captures accurate depth information to distinguish live subjects from 2D photos or video replays. It delivers high security with a face recognition false acceptance rate as low as 0.001% and an overall verification pass rate of 98.85%.

Versatile 3-in-1 recognition modes

The module combines contactless deep-learning face recognition, subcutaneous palm vein scanning, and QR code reading in a single unit. It detects faces at ranges between 30 cm and 120 cm, while palm vein and QR code recognition operate optimally at 15 cm.

Standalone offline edge processing

An onboard SoC featuring a 900 MHz Arm CPU, 600 MHz RISC-V core, and a dedicated 0.5 TOPS NPU executes all neural networks directly on the device. It stores up to 1,000 face templates and 1,000 palm vein profiles locally in 32 MB flash memory, requiring zero computational power from host boards such as an Arduino Uno.

Flexible UART and USB communication

Recognition results stream via a simple 115200 baud UART serial interface for straightforward integration with microcontrollers. The USB connection supports the standard UVC video class protocol, enabling MJPEG video output for live monitoring and video intercom applications.

All-lighting operation and compact footprint

Equipped with 850 nm infrared and 650 nm RGB LEDs, the sensor functions reliably in total darkness as well as non-direct outdoor sunlight. Measuring just 57.8 x 20 x 10.12 mm, it operates on a 5 V to 12 V power supply and includes both Dupont interface and USB cables.

Downloads & documentation

• Product wiki (DFRobot)