Face Recognition

Face Recognition : Matches the Face to that in the database.
Alarms: Option to send Alarms whether Face Matching or non Matching to VMS/ Alarm Center

Applications:

  • Black List Alarm (Terrorists, Shoplifters, Known Offenders)
  • VIP Identification
  • Restricted Zone Security / Lobby Management
  • Face Verification based Access Control & Attendance System
  • Forensic Analysis
  • Viewership Analytics (Face Count, Face Presence, Face Frequency)
  • Demographic Profiling(Age Group and Gender Detection)

Features & Options

  • Works with generic IP cameras with good quality.
  • Works even with PC, you don’t need high end systems
  • Works in both indoor as well as outdoor environments.
  • Works with limited pose changes.
  • Works even when people are walking.
  • Works despite camouflage efforts with spectacles, beard or wig.
  • Works for people of different ethnicity.
  • Face recognition time can be as low as 1 sec.
  • Works for up to tens of thousands of registered faces.
  • Works even with one registered image per face.
Face Recognition

Specifications

  • People walking towards camera with face visible frontally.
  • The face detection allows ±20 degree tilt in both x and y direction.
  • Indoor/outdoor environment with good constant illumination
  • Works at resolutions 640x480 and above, up to 1080p, and for frame rate above 8 fps.
  • Face recognition time is 1-3 sec.
  • Works for a database of up to 10,000 faces.
  • No of registered image per face: 1-10.
  • Last Seen Time of the recognized face is also shown.
  • The environment is indoor as well as outdoor.
  • The camera is able to track the detected face when it is moving and does not generate multiple alarms.

Architecture

  • Face detected from camera is matched against database
  • Alarms are sent to VMS/Alarm Center
Face Recognition Architecture

Camera Installation

  • Cameras should be installed at Angular direction.
  • Camera installation height depends on resolution, typically 1.7 meters.
  • Registration: pixel distance between eyes - min 100 pixels
  • Recognition: pixel distance between eyes - min 100 pixels
  • Detects face typically up to a distance of 3 meters from the camera. The actual distance depends on camera installation height and associated angle of incidence.
  • Constant illumination through out the day (> 200 lux).
  • Detection: pixel distance between eyes – min 60 pixels
Face Recognition Camera Installation

AllGoVision VMS Integration

  • AllGoVision send recognized/non recognized alarms to VMS Viewer application.
  • AllGoVision application can run as a windows service
  • One machine can be used for registration and other machines for recognition
VMS Integration

Software and Hardware

  • OS: Windows Server/8/7/10
  • Database storage: 2 TB
  • Database : MySQL
  • GPU accelerates processing. Nvidia GPU optimization supported
  • NVidia GPU GTX 980 + Core i7 5930 6 core, 32GB RAM, 1TB HDD can support up to 4 Channels

Follow Us