• Embedded biometric module with deep learning facial recognition;
  • Binocular camera;
  • Alive body and mask detection;
  • Age, gender and emotional expressions recognition.

PV50M: Edge Computing Facial Recognition

PV50M is a binocular camera module designed based on high-performance heterogeneous processors. The binocular camera module has built-in facec recognition algorithms. Facial recognition algorithm helps in face detection, alive body detection, mask detection, and can recognize the age, gender, emotional expressions, and identity of a user with his/her face.

These binocular camera modules can be combined with software interfaces to efficiently integrate face recognition functions into the platform according to the application’s needs and platform functions.

It would be the desired solution suitable for time & attendance, access control, and many other fields.

  • 2MP Starlight HDR Dual-Sensor with WDR up to 105dB
  • Detection distance: Facial Recognition - 50 cm to 200 cm.
  • Illuminance of 0.01 lux to 50,000 lux for facial recognition
  • Built-in near-infrared fill light and intelligent switch control based on an image detection algorithm.
  • A thorough SDK available, supporting 3 primary OS (Windows, Android & Linux)
  • Image transmission carried out by the UVC protocol
  • Comparison and results are delivered through HID protocol
GeneralProcessorQuad-core Cortex-47 up to .5G, 1.2 TOPs for INT8
Image SensorDual 1/2.8” HDR CMOS, 2MP resolution, supports RGB+IR
Communication InterfaceUSB 2.0
Communication ProtocolUVC & HID protocol
Power SupplyDC5V 1A (min.)
Operating Temperature-10°C to 50°C
Operating Humidity0 to 90% RH
Dimensions (L*W*H)62 * 38 * 25.3 (mm)
Optical & Image ParametersCamera TypeVisible lightNear-infrared
Field of ViewD=73°; H=65°; V=40°D=73°; H=65°; V=40°
Image Resolution & Frame Rate720*1280@25fps720*1280@25fps
Wide Dynamic Range105dB105dB
Default Output FormatMJPEGNV12
Luminance0.01 lux to 50,000 lux0.01 lux to 20,000 lux
WindowsWindows XP/ Windows 7/ Windows10 (32-bit or 64-bit)
AndroidAndroid 4.1 or above
ZKLiveFace AlgorithmAngle ToleranceYaw ≤±30°, Pitch ≤±30°, Roll ≤±30°
Face Capacity50,000 for 1:1 and 1:N comparison
Face Features Recognition AccuracyAge (±5 y/o) > 85%
Gender > 96%
Emotional expression > 88%
Facial Recognition AccuracyFAR = 0.01% when FRR = 1%, FAR = 0.0001% when FRR = 5%
ZKPalm AlgorithmAngle ToleranceYaw ≤±30°, Pitch ≤±30°, Roll ≤30°, Bend ≤20%
Palm Capacity6,000 for 1:1 and 1:N comparison
Palm Recognition AccuracyFAR=0.001% when FRR=0.68%
CertificationsCE, FCC, RoHS
Recognition ModuleFacial Recognition Module

Data Sheet

Download Date Files
FA50M/FA51M/PV50M Data Sheet 26/04/2021
fa50rn-fa50m-fa51m-pv50m_datasheet.pdf480.82 KB