DIGIFORT Facial Recognition System

Today, technology such as social media, television and other more specialised communication networks play a more influential role in the recognition process. Advancements in artificial intelligence and biometric technology, including the development of Machine Learning capabilities, have led to increased accuracy, accessibility and the widespread use of computerised facial recognition. The significance of this means that facial recognition can occur on an even larger scale and in more challenging environments.

The idea behind all facial recognition technologies is broadly the same: you start with an image of a person’s face (ideally a high quality one, although machine learning means that to a point we can now even use video without reducing accuracy). A fully front facing image is best, think a passport photo, but machine learning and new software has made this more flexible.

An algorithm converts this image into a numeric template, which cannot be converted back to an image and so represents a secure one way system. Every numeric template is different, even if it started out as an image of the same person, although templates from the same person are more similar than templates from different people.

What happens next sounds simple although the technology is extremely complex: templates of people’s faces are taken in real time and compared to those in the database. The technology identifies individuals by matching the numeric template of their face with all the templates saved in a database in a matter of seconds or milliseconds. To put this into perspective, imagine you are at the turnstiles of a busy train station looking for a person on the run.

Digifort facial recognition is designed by modules that can be installed on one or more servers. Modules main ones are:

• Server: API and web interface serves System.
• Analytics Worker: detection and face recognition.
• Streaming Worker: processing video stream from the VMS or direct from cameras.
• Database: Database choice: Postgres (recommended), mysql

In the most general case Digifort facial recognition operates as a distributed architecture that streams video of cameras are processed by Monitoring Servers installed on site. These servers monitoring modules are composed of management of the video stream (streaming workers) and a analysis module (module analytics). Additionally a central server contains the platform main and common display interface and configuration. The information registered individuals and events generated can be stored in a common database between different locations.

The primary requirement regarding the specifications of the camera is the minimum size (in pixels) face captured. To meet this requirement two factors come into play:

• Hardware Specifications Camera: Resolution, fps (frames per second) and type sensor.
• Camera positioning relative to the area of interest to be monitored.
• Minimum face size: 96 x 96 pixels.
• Minimum resolution: 1080p camera and 2MP (2 megapixels).
• Sensor: minimum size of 1/3 ” (progressive), RGB (color) minimum capture frequency: 10 fps

Server type A, you are responsible for running the web interface and visualization and configuration coordinate monitoring servers besides the management database of individuals and events. The technical specifications are:

• Core i7-8700K 4.2GHz Xeon
• 32GB (4 x 8GB) DDR4-3200 Memory
• 10TB 3.5 “7200RPM Hard Drive Internal
• Linux OS Ubuntu 16.04

One or more servers monitor (type B) will be installed at each location, depending on the number of cameras to monitor.
These servers require great computing capacity so include dedicated graphics cards. The technical specifications of these
servers are summarized in the following table:

Synchronization database shared between locations and real-time notifications require monitoring servers count with a broadband connection standard between different locations. Requiring notifications outside the internal network, you must have Internet connectivity to the outside.

Buying the license Digifort facial recognition, it has included support configuration and maintenance facial recognition software and database of individuals for the period of 1 (one) year.

The system can be calibrated to operate at various operating points, privileging rate positive hits (number of detections) or a lower rate of false positives (false alerts). The system, with typical parameters, has a positive recognition rate of 98% with a rate false positive 0.067%

The system allows registration of individuals through different mechanisms

• Previously stored images of the individual, which are uploaded to the web interface system for registration.
• Mobile applications for Android and IOS allowing the system administrator or person authorized, register a person capturing a sequence of photos of the face.
• Forensics video files. The system allows you to upload a video, detect all individuals present and semi-supervised manner, registering new subjects of interest.
• Auto enrollment of users through the web interface and a standard USB camera (from good quality) connected.
• A feature of the system is that enroll people from previous detections unknown individuals. The system has the ability to record the faces of individuals not stored in the database to be subsequently selected by the user and registered.
• The facial recognition system allows individuals to register in a gallery whitelist (Allowed users) or gallery blacklist (unwanted or unauthorized subjects) with different for each notification. These lists can be configured for each point and monitoring is the ability to generate calendars days and hours allowed for each individual.
• Updating the database, which includes registration, removal or renovation of individuals and Instant registered is performed for each of the locations and monitoring points way.

• Reports of these events can be downloaded in PDF format.
• Add / delete / modify cameras
• Add / edit / delete individual enrollees
• Configuring notifications
• Enlistment by Web camera, images or video