AUREUS 3D DELIVERS FAST, ACCURATE, SCALABLE AND COST EFFECTIVE FACIAL RECOGNITION Software

In 1966 a mathematician and computer scientist named Woody Bledsoe was assigned by a US intelligence agency to find a way for a computer to utilize facial recognition. The problem was, given a large database of images (a book of mug shots) and a photograph (another mug-shot), it was challenging to determine a match. Fifty years later, facial recognition still refers to a computer system that matches two images of faces. The difference is we’re now searching databases scaling to hundreds of millions of images, and the “photograph “is now a live video stream.

CyberExtruder approaches the problem of matching faces from a vastly different perspective –3 dimensions (3D). The approach is unique enough to be granted patents and acknowledged by National Institute of Standards and Technology (NIST) as providing increased results over conventional (2D) matching efforts.

investigating facial recognition technology?

If you are investigating facial recognition technology as the means to solve an identity management problem, we encourage you to dig in and get to know us. It’s easy to say you’re the best, the fastest, the most accurate or the industry leader, but it’s another thing to be able to prove it by solving your customer’s challenges and delivering ROI.

CyberExtruder has earned its reputation for having the best technology and utilizing a state-of-the-art facial recognition algorithm and being highly responsive to client needs and delivered on time and within budget. Whether you’re an OEM Partner, a Solutions Integrator or an End User, we have the facial recognition software solutions to meet the challenge.


CyberExtruder Aureus 3D Facial Recognition Product Brochure PDF downloadCyberExtruder Aureus 3D Facial Recognition Technical Specifications PDF download


Aureus Facial Recognition software In use

Aureus 3Dv

Aureus 3Dv (v for Video) is an easy to install and simple to use face matching application designed specifically for use in live, uncontrolled environments.

Aureus 3Dv is the world’s most advanced 3D Facial Recognition Software for use with conventional video and images, and featuring a fully integrated 3D reconstruction, pose correction, expression, and illumination neutralization toolset. Aureus 3Dv mitigates the real-world challenges to facial recognition created when security cameras are mounted in less-than-ideal locations.

patented 3D modeling

Aureus 3Dv gains its facial recognition search advantage through its patented 3D modeling, sophisticated computer vision algorithms and trained machine intelligence that:

  • Identifies human subjects at a distance where the subject’s face is as little as 25 pixels high
  • Performs analysis and recovery of camera focus, head pose, quality and directions of light sources, facial expression, and security camera lens distortion
  • Executes 25 million template comparisons per second per 3.3GHz CPU
  • Provides Operator assisted gallery boosting to further refine existing galleries.

More detailed information on the performance of the Aureus platform can be found on our facial recognition technology page

Aureus 3Dv compensates for poorly positioned cameras that would render most facial recognition systems useless. Lines of sight which result in captured facial images turned away from the camera as much as 70 degrees side to side or 25 degrees up and down are smoothly transformed into 3D, reconstructed and pose, expression and illumination neutralized for optimal matching.

multiple frame analysis

The 3Dv performs multiple frame analysis refining the identity assessment of a subject as new video frames are consumed and added to each tracked subject.

Aureus 3Dv ‘learns’ from subjects in motion when they travel through variations in illumination producing a more sophisticated comparison to the existing gallery.

Designed specifically to monitor video feeds, Aureus 3Dv provides an automated identity processing hub for organizations with identity management security requirements. Most amazing of all, Aureus 3Dv matches faces at a rate of over 25 million per CPU per second so it doesn’t even break a sweat given galleries of tens of millions of individuals.

Read more

+

Aureus SDK Facial Recognition Development

Aureus 3D SDK

Aureus 3D SDK (Software Developer’s Kit) contains all the code libraries you need to efficiently assemble a custom facial recognition system to meet your unique requirements.

The SDK includes a comprehensive set of software modules which can be used by our partners to build and integrate a particular facial recognition capability. Easy to install and integrate, the Aureus 3D SDK provides the fastest facial recognition software available.

Base performance

On a base system with a 3.3GHz CPU, Aureus 3D generates a facial template in 100 milliseconds – nearly twice as fast as the nearest competitor and compares two templates in 0.040 microseconds – that’s 25 million comparisons per second per CPU. Aureus 3D was designed from the ground up using industry standard protocols and development environments. Our facial recognition engine works with any IP or USB video camera, or archival video file, and conveniently communicates with third party applications via XML.

Aureus 3D overcomes the challenges of subject pose, expression, partial facial occlusions and environmental lighting, and optimizes the competitive advantage of 3D over 2D technology.

The 3D helps you deliver fast, scalable facial recognition systems to meet your customers’ current and evolving needs, and real return on your customers’ investment. Our patented 2D to 3D pose correction technology—with deep convolutional neural network algorithms—offers exceptional performance and meets real-world challenges such as lighting, pose, occlusions, motion, crowds, and expression.

With industry-leading speed and a 128-byte template, size is the absolute smallest of any facial recognition algorithm anywhere. Easily deployed, Aureus 3D’s industry-leading speed, match accuracy, and lower cost enables speed to market with operationally viable facial recognition solutions.

What’s in the Aureus 3D SDK

The Aureus 3D SDK is a facial recognition development package which has been written in Microsoft’s Visual C++ and contains fully documented example projects to quickly get you up and running and building your custom facial recognition applications. There are also test datasets and sample video files to assist you in tuning your applications as you work.

Features

  • Automatic Feature (Eye, Nose, Mouth) Detection
  • Automatic Face Detection
  • Face Tracking
  • Automatic Head and Shoulder Detection
  • Real-time video capable at 30 fps
  • 1:1 Matching
  • 1:n Matching

Matching Engine Robustness

  • Pose capable to 60° Yaw, 20° Pitch, 30° Roll
  • Handles partial facial occlusions
  • Handles eyeglasses
  • Bypasses beards and hairstyles
  • Handles large expressions
  • Manages most sunglasses
  • Manages irregular lighting across the face
  • 29 pixel Minimum distance between the eyes
  • 32 Minimum grayscales within the face

Template Specs

  • Template size is 128 bytes
  • Template generation speed is 100 milliseconds
  • 25 million template comparisons per second per CPU
  • Template generation is re-entrant thread safe
  • Template matching re-entrant thread safe

Host Platforms

  • Windows 64 bit
  • Windows 32 bit
  • Ubuntu Linux
  • SQ Lite for database management

Read more

+

Gentleman Groper Captured by Aureus Facial Recognition software

Aureus 3Df

Aureus 3Df (f for Forensic) is a suite of ready to use tools designed specifically to meet the biometric identification demands of law enforcement and intelligence investigators.

If you’re currently using facial recognition as an investigative tool, you know how critical it is to have a high quality, front facing images of the suspects you’re attempting to find. You also know that criminals never stop to pose for their mug shots during the crime. When your only image evidence is inadequate quality where can you turn?

Forensic Facial Recognition

Aureus 3Df is the only suite available to investigators who need to overcome lighting, pose and resolution challenges. Typically, images from security systems are very low resolution and as a result, usually produce poor matching results.

The “Gentleman Groper” Case

Gentleman Groper Captured by CyberExtruder Facial Recognition Technology

The pictures above are from a New York Police Department case file in which this individual was seen numerous times on surveillance cameras, but the quality of the facial images were not sufficient by themselves to allow investigators to match them against known suspects. These pictures typify the quality of image evidence encountered by Police.

Seen below, Police commissioner Ray Kelly holding up a surveillance image that was frustrating the NYPD’s ability to identify the suspect. Using the facial images seen above, a composite image is created.

Gentleman Groper Captured in NYC by CyberExtruder Facial Recognition Technology

CyberExtruder’s all new Aureus 3Df delivers quick and easy 2D to 3D conversions of images and video files into actionable mugshots and probe images – and best of all, Aureus 3D extends the working parameters of facial recognition.

Read more

+

Aureus Image Analysis SDK for Determining Image viability in Facial Recognition Applications

Aureus Image Analysis SDK

CyberExtruder provides an image analysis SDK written in C and C++ which consists of a Dynamically Linked Library (DLL), and two sample applications (including source code) which demonstrate how to use the SDK. The SDK provides functionality to handle:

  • SDK DLL loading and function pointer wrapping
  • SDK initialization, license confirmation, get last error, general SDK support functions
  • Basic image handling: load, create/copy, free
  • Single image analysis handling
  • Batch image analysis
  • Video image analysis handling
  • Configurability of ISO test thresholds

Comprehesive Image Checking

The Image Analysis SDK provides a comprehensive check of images by implementing a series of ISO tests. Each test is one factor in determining an image’s suitability for use in facial recognition (FR) applications.

Our Analysis SDK provides image analysis in a series of use patterns:

  • Command line analysis of single, list, or directories of images to generate reports
  • Interactive applications performing controlled image analysis
  • Analysis of images on disk, in memory, and in video streams
  • Receive back a detailed Image Analysis Results structure

In all cases, the purpose is to accept facial imagery and generate a report describing that image’s suitability for use in ISO certified FR applications.

TESTS FOR COMPLIANCE WITH ISO 19794-5 SPECIFICATION

  • Single face confirmation
  • Pose detection for rotation, tilt, and roll
  • Horizontal position confirmation
  • Vertical position confirmation
  • Eye detection
  • Uniform lighting measurement
  • Shadow detection
  • Sharpness estimation
  • Eyeglasses detection
  • Tinted glasses detection
  • Exposure estimation
  • Head size and position detection
  • Unnatural skin tone detection
  • Red eye detection
  • Uniform background detection
  • Hot spots detection
  • Format information extraction
  • Grayscale and color image support
  • Closed eyes detection
  • Closed mouth detection

Read more

+