Integration: RecFaces
Integration Capabilities
Id-Me Face Recognition Module
MULTIMODAL BIOMETRIC by
Enterprise Video Management System
Open platform flexibility
Ultimate Reliability
Software & Hardware installations
+
Safe and Secure
Convenient and Reliable
Features
Human face detection in video streams. Creation and maintenance of a visitor database.
Notification of visitor presence from a pre-defined visitor list.
Notification of unauthorized visitor presence in a high security area.
Fast database search (“smart search”) of a suspect by their face image, with an option to
extract the following attributes: video camera id, location id, time and date.
Visitor identification via the mobile app.
The possibility to create smart business solutions through integration into ERP/CRM
systems via the use of open-source API.
The Id-Me Face Recognition Module is a software-hardware solution capable of face
detection via video stream processing. Subsequently, it is also capable of face search
through matching to a pre-defined database or a list. From the moment of detecting a face
image of sufficient quality, the matching to a database is completed in under 1 second. The
system supports visitor categorization into groups, such as “staff”, “contractor”, “wanted” etc.,
which allows the structuring of notifications. Each time a person is captured by the system,
the following data record is stored in the system database: the captured person’s face crop,
the respective video frame this image has been cropped from, the person’s biometric
template, as well as the date, the time, and the camera id, along with some other attributes
necessary to make database search and the search result filtering faster and more
convenient. Hence, the system only needs a few seconds for accomplishing tasks a human
operator would otherwise need minutes or even hours for. The face recognition system also
has an open-source API, which simplifies the integration with external IT systems, the usage
of recognition results for business intelligence purposes, and the creation of an integrated
infrastructure aimed at improving business safety and efficiency.
Improved Data Storage Efficiency
For continuous 1080p H.264 video recording at 25 frames per second, 16 GB of storage
space per each video surveillance channel per day would normally be required.
Each database record of the Id-Me Face Recognition System, containing the captured
person’s cropped face image, the respective full video frame, and the biometric template,
requires only c. 120-150 kB. Hence, when using the Id-Me Face Recognition System, the
same 16 GB of HDD space would suffice for recording data on c. 140,000 captured visitors,
which equals 280 days of system operation at 500 visitors per day.
In a typical video surveillance system, the video archive would normally be overwritten with
new footage every month. In contrast, the Id-Me Face Recognition System data is stored in
an indexed database, which may be available after as long as a year or more.
Multimodal Biometrical Platform
Efficient face detection and identification
Long-term information storage
Convenient Smart Face Search
Tags:
Analytics
Smart search
Improved data storage
Fast Search of Wanted Persons
Manual search of a person in footage generated by 10 video cameras during just one day
could take several man-hours. This task is especially difficult when a person is unfamiliar to the
security officer.
The smart biometry-based search function provided by the Id-Me Face Recognition System
requires just an image of the person to succeed. In only a few seconds, the system will
perform a database search and identify all locations where the person was captured by the
system within a specified time frame. Modern biometric systems detect persons with a much
higher efficiency than human operators, especially when a person is unfamiliar to the operator.
The security surveillance data processing speed may be a matter of life or death. Improved
search efficiency of wanted persons greatly increases the overall efficiency of a security
system.
Immediate Notification of Operators
Security operator’s level of concentration may decline through the work shift, which may
become the critical point of failure in safety, security or incident prevention.
The Face Recognition System generates automatic notifications (alarms) as soon as either an
unauthorized / unknown person (i.e. not in white list), or a person present in a black list, is
detected. Notifications can be displayed at the security operator’s desk, on remote mobile
devices, or both.
This ensures the security operator is fully aware of the presence of an unauthorized person in
a security area in a swift and timely manner.
Business Analytics
In addition to increasing security systems performance, biometric identification systems can
be effective for solving business-related problems, while the open-source API available for the
Id-Me Face Recognition System also makes it easy to integrate with external IT systems (e.g.
ERP and CRM).
- Big Data: one step ahead of the competition. Collecting and structuring customer data in
retail and banking in order to improve communications and the quality of service.
- Upsale and effective communication. Improved interaction between the staff and the
customers through notifying the staff about the customers shopping preferences (available
from the CRM).
- Business process automation. Customers can be identified at the point of entering an
electronic queue, at which point their information can be sent to and processed in CRM with
the possibility of notifying a senior member of staff about the arrival of a VIP-customer.
- Simplified attendance recording. Staff arrival and departure times can be automatically
recorded through integration of the Id-Me Face Recognition System with external IT systems.
Staff will be able to interact with customers in a targeted manner, enhanced by the information
about the frequency of visits, purchase history, and the results of prior communications and
interactions, which are recorded in the CRM system.
Integration: RecFaces
The system fully supports Id-Me Light Bio Core & Video
10,000 photos.
Small systems in which identification of the face image on one
with the reference photo database.
The system consists of:
Server is 10 servers.
The Enterprise Id-Me for VX class equipment is based on
unlimited.
System architecture of solution
Architecture of the biometric Id-Me module within the Pelco VX.
Types of Id-Me solutions for Pelco VX
Type Description Typical use
Light
Id-Me for VX
Professional
Id-Me for VX
Enterprise
Id-Me for VX
Stream Processing Server on one server, processing the
single FULL HD video stream of up to 25 fps.
The limitation on the reference photo database for this solution is
-
Central server used for identification, data and configuration
storage - Id-Me Professional Bio Core Server.
- Backup server (optional) Professional Bio Core Redundancy
Server.
- Servers processing up to 2 video streams in FULL HD, up to
25 fps - Professional Video Stream Processing Server.
The limitation on the reference image database for this solution
is 100,000 photos per Professional Bio
Limit on the number of Professional Video Stream Processing
Servers that can be connected to one Professional Bio Core
specialized high-performance DELL servers. This ensures
cost-optimization of server hardware in large installations.
Calculation of cost for the server hardware and licensing in
Enterprise Id-Me for VX systems is available on request
from Pelco by Schneider Electric.
Reference photo database limit for this solution is customized
on request.
The number of simultaneously connected video cameras i
Core Server.
camera is required.
If necessary, several territory distributed Id-Me Light Bio Core
& Video Stream Processing Servers can be installed in sync
Territorially-distributed mid-level systems, in which
simultaneous identification is required from a large number of
video cameras with a limited reference photo database.
If necessary, several Professional Bio Core Servers can be
installed in sync with the reference image database and the
required number of Professional Video Stream Processing
Servers to process the required number of video stre
Corporate-class territorially-distributed systems where
simultaneous identification is required on a large number of
video cameras, without restrictions on the size of the
reference image database.
s
ams.
Integration: RecFaces
The server performs identification function, sends
format with up to 25 fps.
Biometric identification: matching procedure, data storage, system
detection and metadata storage.
Pelco VX Infrastructure Servers
Pelco VX Management, storage server infrastructure.
Operation and data storage of the CCTV system.
Biometric-system integrated VX plugins:
- Biometric check and identification services.
A server that processes up to two Full HD video
frame, biometric template.
The server performs identification functions, sends
Processing Servers.
Professional Bio Core
Redundancy Server
Second Bio Core Redundancy Server for mission-critical
facilities.
Pelco VX Infrastructure Servers
Pelco VX Management, storage servers infrastructure.
CCVT-system management and data storage infrastructure.
Biometric-system integrated VX plugins:
- Biometric verification & identification services.
Description of the hardware Light Id-Me for VX
Title Description Purpose
IP-Camera Network camera managed by the Pelco VideoXpert.
Light Bio Core & Single Video
Stream Processing Server
Integrated VX Plugins
notifications to Pelco VX, stores identification results.
The reference base of photo-images is up to 10,000. One
video stream from an IP camera is processed in FULL HD
- Notifications (blacklists, whitelists etc.);
- Detected face archive;
- Smart face database search;
Forming a video stream for processing on Light Bio Core & Single
Video Stream Processing Server.
configuration, VX integration services.
Single video channel: face detection, quality control, biometric
template creation, sending data for matching procedure, face
Biometric graphical user interfaces.
Description of Professional Id-Me for VX technical tools
Title Description Purpose
IP-Camera Network camera managed by the Pelco VideoXpert.
Professional Video Stream
Processing Server
streams with a frequency of up to 25 fps per channel.
The server also performs the functions of data storage:
detected person, date, time, camera ID, contextual
Forming a video stream for processing on the Professional Video
Stream Processing Server.
Face detection, quality control, biometric template creation, sending
data to Bio Core Server for matching procedure and metadata
storage.
Professional Bio Core Server
Integrated VX Plugins
notifications to Pelco VX, stores identification results. The
reference base of photo-images is up to 100,000.
Simultaneous work with up to 10 Video Stream
- Notifications (black, white, etc. lists);
- Detected faces archive;
- Smart face archive search;
Biometric identification: matching procedure, data storage, system
configuration, VX integration services.
Redundancy Server.
Biometric graphical user interfaces.
Integration: RecFaces
Camera resolution
1080p (1920x1080)
or higher
Video stream frame-rate
15 fps
or higher
Wide dynamic range
Preferable
Lens, focal length
f= 15-50, Megapixel, Auto Iris
or higher
The camera should be well focused so that the face is sharp, of high contrast, clearly visible and
discernible by naked eye.
maximum of 600 Lux).
Recommended inter-eye-distance for enrollment is 120 pix. or more.
Recommended inter-eye-distance for identification is 80 pix. or more.
Video camera should be firmly fixed on the ceiling with a special bracket to minimize blur caused
by camera movement.
The recommended location of the camera: the object looks and moves forward in the direction of
the camera, the object moves along the line of sight of the camera.
start identifying individuals located further than 8.0-8.5 m.
Optimal field of view
Preferable camera field of view width is 2 m wider than the distance of the straightened hand.
Eye-glasses
Clear eye-glasses are available.
Recommended IP-Camera
A different camera may be used to obtain a
operating conditions of the system".
A different lens may be used to obtain a video
conditions of the system".
Light Bio Core & Single Video Stream Processing Server parameters
- up to 1 FULL HD Channel processing
i7-7700T(2.9 MHz, 8M, QC), 32GB (2x16GB) 2400MHz DDR4, 3Y
NBD
HDD 1
System disc: SSD Intel Original SATA III 480Gb 540s Series M.2
or higher
HDD 2
History storage disc: SSD SATAIII 1024Gb, 2,5”
or higher
Operation system
Windows 10 Professional
or higher
Professional Video Stream Processing Server:
- processing up to 25fps per channel
NBD
HDD 1
System disc: SSD Intel Original SATA III 480Gb 540s Series M.2
or higher
HDD 2
History storage disc: SSD SATAIII 1024Gb, 2,5”
or higher
Operation system
Windows 10 Professional
or higher
Professional Bio Core Server:
- up to 10 Professional Video Stream Processing Server connected
i7-7700T(2.9 MHz, 8M, QC), 32GB (2x16GB) 2400MHz DDR4, 3Y
NBD
HDD 1
System disc: SSD Intel Original SATA III 480Gb 540s Series M.2
or higher
HDD 2
History storage disc: SSD SATAIII 1024Gb, 2,5”
or higher
Operation system
Windows 10 Professional
or higher
Professional Bio Core Redundancy Server
i7-7700T(2.9 MHz, 8M, QC), 32GB (2x16GB) 2400MHz DDR4, 3Y
NBD
HDD 1
System disc: SSD Intel Original SATA III 480Gb 540s Series M.2
or higher
HDD 2
History storage disc: SSD SATAIII 1024Gb, 2,5”
or higher
Operation system
Windows 10 Professional
or higher
Recommended equipment and parameters of Light and Professional servers
Title Description Comments
IXE22 or higher Sarix Enhanced SureVision3 Full HD camera
YV3.3x15SR or higher Fujinon Lens System 15-50 mm., F1.5
- up to 10 000 enrolled faces in DB
Dell OptiPlex 7050 micro
- 2 FullHD cameras connected
Dell OptiPlex 7050 micro
- up to 100 000 enrolled faces in DB
Dell OptiPlex 7050 micro
Dell OptiPlex 7050 micro
i7-7700T(2.9 MHz, 8M, QC), 8GB (2x4GB) 2400MHz DDR4, 3Y
video stream, as long as it meets the
requirements of Table "Recommendations on
stream, as long as it meets the requirements of
Table "Recommendations on operating
or higher
or higher
or higher
or higher
Recommended operating conditions
Recommendations on operating conditions of the system.
Parameter Value Comments
Focus of the camera
Uniform and constant light level should be provided indoors. For optimal face recognition indirect
Illumination
Face posture for enrollment and
identification
Inter-eye-distance
Camera movement
Camera location
Optimal camera installation height
lighting should be such that those objects had a uniform illumination without shadows or glare.
The recommended light intensity should be about 300 Lux (a minimum of 150 Lux and a
Recommended head
Recommended head yaw for identification (Y, Figure 1) is ±15 degrees from frontal position.
Recommended head roll for enrollment (R, Figure 1) is ±7 degrees from frontal position.
Recommended head roll for identification (R, Figure 1) is ±15 degrees from frontal position.
Recommended head t for enrollment (P, Figure 1) is ±
Recommended head
Optimal camera installation height above the floor is between 1.5 and 2.2 m. It is desirable to
yaw for enrollment (Y, Figure 1) is ±7 degrees from frontal position.
7 degrees from frontal position.
pitch for identification (P, Figure 1) is ±10 degrees from frontal position.
Integration: RecFaces
Recommendations for video camera placement.
Entrance
Derection of movements
Zone of face recognition
Zone of receiving more than 20 frames
8.5 m
As far as possible
15
to 10
1.4 m
4.5 m
Data Storage in Id-Me Light and Id-Me Professional
The Id-Me for VX has a data storage function, which aids in swift investigation of security
incidents. This function makes it possible to find a person by his/her face image, and to
determine when and where this person was captured by the system within a specified
time frame - using either the video footage archive (when not yet overwritten by the VX
system), or the database records of the Id-Me system. The system uses the following
servers: Light Bio Core & Single Video Stream Processing Server, Professional Video
Stream Processing Server, Professional Bio Core Server and Professional Bio Core
Redundancy Server for storing the following data: the captured person’s cropped face
image, the respective full video frame, both in either JPEG or PNG format,, and the
biometric template, requires only c. 120-150 kB. It is also possible to store the captured
person’s cropped face image and the biometric template that requires only c. 50 kB. A
data record, which is added every time a person is captured by the system, is from c.
50 to 150 kB in size and depends on configured data-model in database.
When using the Light Bio Core & Single Video Stream Processing Server, Professional
Video Stream Processing Server, Professional Bio Core Server and the Professional Bio
Core Redundancy Server, the HDD1 will be used as the system drive only. If the HDD2
is then used for storage of the Id-Me database, and if it has 1024 Gb of storage space
available, it will be sufficient for storing over 8 million records in extended data-model
(respective full video frame, cropped face image, biometric template, metadata) or over
19 million records in standard data-model (cropped face image, biometric template,
metadata) each covering a single event when a person was captured by the system,
and each instantly available for running security incidents investigations or for utilizing
business intelligence-related functions - all on the same server. Thus, knowing the visitor
throughput at the focal location, the archive timespan which can be covered by the
Id-Me system can be readily estimated.
Roof
As low as
possible
0.8 m
(example)
2.2 m
(example)
Floor
When the data is received from two video cameras simultaneously using the Professional
Video Stream Processing Server, each camera will generate an individual database
record, both of which should be accounted for in estimating the archive timespan.
When using Professional Id-Me for VX, we recommend recording operational data onto
the Video Stream Processing Servers, while data records related to high-priority
individuals (e.g. blacklisted individuals, VIPs, or unauthorized persons moving to/from
security areas outside normal working hours etc.) should be stored on the Professional
Bio Core Server. This approach ensures higher longevity of high-priority data.
For the Light Id-Me for VX system, the operational data is recorded onto the HHD2 of the
Light Bio Core & Single Video Stream Processing Server.
Figure 1. Yaw, Pitch, Roll angles
About RecFaces
The company's product is a Id-Me multimodal biometric identification platform
that allows using the built-in identification algorithms to receive, store and
process various biometric features of each person, such as face image,
three-dimensional face image, vein pattern, etc.
↘
Contact the sales representative in your
area today, or visit
id-xpert.com
to learn
more.
Loading...