Parameter

Body Temperature Detection Specifications

Specifications ( GL Exclusive Feature)

Read more About ZKTeco F18 in BD

Introduction of deep unconstrained face verification and face anti-spoofing technology

ZKTeco’s pioneering Deep Unconstrained Face Verication and Face
Anti-Spoong Technology are the new milestones representing the historically
new level of facial recognition technology, by enabling learning of regularized
metric to compare facial features of human face to achieve higher accuracy
than the traditional method, and precise pixel depth information for high
accuracy against spoong by dierentiating between a face and a at shape.
ZKTeco is always striving for advancement of technologies to oer you better
life. Deep Unconstrained Face Verication and Face Anti-Spoong Technology
are your best choice of face verication solutions.
Introduction of deep unconstrained face verication
and face anti-spoong technology
Taking advantage of using MetricNet, it consistently achieves higher
accuracy than the traditional method, and achieves a higher
accuracy from 98.06% to 99.80% by learning image metric and
comparison with facial features, as a result, the discrimination ability
of the network is signicantly increased.
Meanwhile, it enables the network to learn the best metric and follow
the target distributions for improved accuracy and results.
Advantages of using Deep Unconstrained Face Verication
Introduction of Deep Unconstrained Face Verication
and face anti-spoong technology
Deep unconstrained face verication
What Deep Unconstrained Face Verication is
In recent years, facial recognition has been widely used in various industries with its high
accuracy and convenience, huge progress has been made for deep learning and facial
recognition. Today, the new algorithm has been developed for higher security.
•Deep Unconstrained Face Verication
•Face Anti-spoong Technology
Deep unconstrained face verication plays a critical role in the
learning process of regularized metric to compare facial features of
human face. Dierently from the traditional method such as FaceNet,
it nds facial the best features and therefore shapes the decision
space by learning a latent representation in which matching and
non-matching pairs are mapped onto clearly separated and
well-behaved target distributions.
In particular, the network jointly learns the best feature
representation, and the best metric that follows the target
distributions to be used to discriminate face images.
Forehead
Brows
Eyes Mouth
Nose
Cheeks
Deep unconstrained face verication Face anti-spoong technology
How Deep Unconstrained Face Verication works
After input the image of the human face,
FeatureNet are able to extract the most distinctive
facial features from the input pairs. The
architectural design of FeatureNet is critical as it is
able to employ any state-of-the-art neural network
architecture able to learn good features.
• Step1
Feature net
The feature tensors of the human face are linked
together and concatenated and then passed to the
MetricNet.
• Step2
Concat
MetricNet map the input pairs of the image onto
target distributions in the latent space, in these
step, it is able to learn the best metric based on
feature tensor and to map it onto the target
distributions.
• Step3
MetricNet
The best metric follows the target distributions, to
be used to discriminate face images. Matching
pairs (same user – blue) are mapped to a target
distribution whose mean value is far from that of
the non-matching pairs (dierent users – red).
• Step4
Target distribution Target Distributions
Non-Matching
Pairs
Concat
Input
Image1
Input
Image2
FeatureNet
Step1
Step2
Step3
Step4
MetricNet
Matching
Pairs
Dierently from the traditional method of Residual
Block, the latest technology of Residual Scale
Attention Block（RSAN） has a higher security for
spoong face in input image.
Face anti-spoong technology advantages
Face Anti-Spoong Technology
What Face Anti-Spoong Technology is
Biometric system is generally vulnerable to presentation attacks, typical example of physical presentation attacks
include print, video, 3D masks and makeup. Face anti-spoong (FAS) plays a vital role in securing the face recognition
systems, its higher detection algorithm secures face recognition system and attracts more attention from various
industries.
The latest Face anti-spoong technology adopt the
most reliable anti-spoong technique uses a 3D
camera. Precise pixel depth information provides
high accuracy against presentation attacks, the
dierence between a face and a at shape can be
detected.
While 3D attacks still cause diculties for facial
recognition, its stable performance makes this
anti-spoong the most promising.
How face anti-spoong works
It extracts shallow features, depicts cheeks, edges, and
shapes of facial features.
• Step 1 Residual Block
It concats low-level and multi-scale feature maps.
• Step 2 Multi-Scale Module
It was useful for retaining local features, enhancing
global representing capacity.
All hair, face, and background regions have a relatively
strong activation for the living faces while the facial
areas contribute weakly for the spoong faces.
• Step 3 Attention Module
It forms a multi-dimensional feature “Tensor” that is
mixed together after intra-level and inter-level
aggregation, containing both low-level spatial
information and high-level discriminant.
It describes the dierence between the edges and
shapes of facial features between living bodies in
dierent scales versus spoong faces and depicts the
semantics’ dierence between the living body and
spoof body.
• Step 4 Feature Fusion
Distinguishing living faces from spoong faces, feature
maps with discrimination ability relying on decision
mapping transfer to a specic range. If the result is
higher than the standard value, the sample will be
regarded as a living face; if the result is below the
standard value, it will be considered fake-face, which
means access is denied.
• Step 5 Decision Mapping
Residual Block
Multi-Scale Module
Attention Module
Feature Fusion
Decision Mapping
Facial Recognition Acceptance and False Rate
The newest Facial Recognition has a significantly enhanced recognition speed
– Actual facial recognition speed after installing firmware: 0.3s
– Facial recognition algorithm speed: 0.15s (speed doubled compared with the past version)
Version/Type FRR
(False Rejection Rate)
FAR
(False Acceptance Rate)
East Asia
(e.g. China, Korea, Japan) 0.01% 0.02%
Europe and the United States
(e.g. Europe, Russia, North America) 0.19% 0.03%
Latin America
(e.g. Mexico, Peru, Chile) 0.25% 0.04%
South Asia
(e.g. India, Thailand, Vietnam) 0.29% 0.04%
Middle East
(e.g. Dubai, Saudi Arabia, Egypt) 0.33% 0.04%
Africa
(e.g. South Africa) 0.55% 0.05%
Diverse product oerings
ZKTeco oers a wide range of high-quality products that meet the needs of global customers, from facial recognition
terminals, network cameras to metal detectors that all integrate with the facial recognition technology. With
emphasis on both quality, technology, and cost-eectiveness, ZKTeco oers the best solution in a wide range of
dimensions.
Elite Access
Multi-Biometric T&A
and A&C Terminal
• Facial Recognition
• Palm Recognition
Facedepot 7B
Multi-Biometric T&A
and A&C Terminal
• Facial Recognition
• Fingerprint
• Palm Recognition
• RFID
Multi-Biometric T&A
and A&C Terminal
• Facial Recognition
• Palm Recognition
Proface X Multi-Biometric T&A
and A&C Terminal
• Facial Recognition
• Fingerprint
• Palm Recognition
• RFID
G4L
Multi-Biometric T&A
and A&C Terminal
• Facial Recognition
• Fingerprint
• Palm Recognition
• RFID
SpeedFace V5L
Multi-Biometric T&A
and A&C Terminal
• Facial Recognition
• Palm Recognition
• RFID
SpeedFace V4L
ZKTeco’s touchless biometric solution is a good t for this situation that is preventing people or patients from touching
the door handle. The solution has been widely used in many practical scenarios, including hospitals, educational
institutes, factories, construction sites, shopping malls, IT parks, public transportation, banks, business organizations,
small to medium enterprises, government organizations and so on.
!”#$%&'( #)
!
Hospitals Educational Institutes Factories
Construction Sites Shopping Malls IT Parks
Public Transportation Business Organizations Small to Medium Enterprises
Banks Government Organizations

Notification about Stopping ZKTime Web 2.0 Maintenance and License Sales

With the expansion and development of the company’s business, the new version of the overseas time and attendance software BioTime 8.0 has been released. In response to the company’s development strategy and providing customers a better product experience, ZKTime Web 2.0 will be discontinued from 1st,October 2019. The specific notifications are as following:

1.From 1st,October 2019, all regions except Chile will stop ZKTime Web 2.0 license sales.

2.From 1st,October 2019, the standard version of ZKTime Web 2.0 will no longer be accepted for customization.

3. For customers who have purchased the mainstream version of ZKTime Web2.0, if the customer is willing to switch to BioTime 8.0 or there are mainstream bugs while using ZKTime Web2.0, it’s supported to migrateZKTime Web2.0 specified data to BioTime 8.0. For the BioTime 8.0 license, please contact with corresponding license provider for license policy.

4. For the customized version of ZKTime Web2.0, if there is a bug in the software during use, and want to migrate customized function and data to BioTime 8.0, please contact product manager Nancy for estimation.

5. For the ZKTime Web 2.0 that has already been sold, the after-sales service will continue to be provided, and the maintenance for all version of ZKTime Web 2.0 will be completely stopped on 31st September, 2020.

Guidelines on the Use of Biometric Data

Biometric Data User refers to any person who collects or use biometric data utilizing ZKTeco’s device, hardware and software

1. The purpose of biometric data collection

1.1. Use of biometric data & related activities

It is data user’s responsibility to inform all users or persons whose biometric data are collected, that all the collected biometric data are only used for time attendance and access control purposes.

1.2. Lawful and fair means

Data User must collect users’ biometric templates only on a legal and fair basis.

1.3. Adequate but not excessive

ZKTeco’s devices and all biometric templates stored in its management software are built with precise algorithms and may be in various methods encrypted, in order to prevent data leakage of users to the greatest extent.

2. Accuracy and Duration of Retention

2.1. Accuracy of personal data held

ZKTeco, with rich experience of software and hardware development, to the greatest extent ensures that all biometric data and personal data are stored accurately, in order to provide customers best user experience.

2.2. Personal data not being kept longer than is necessary

When using ZKTeco’s software and hardware, data user is responsible for ensuring that all users’ data are not stored in ZKTeco’s software and hardware unless necessary.

2.3. Prevent any personal data transferred to the data processor from being kept longer than necessary

ZKTeco’s hardware applies different biometric templates during operation, but all templates are not remained in the CPU of the hardware, and after users’ biometric templates are deleted in ZKTeco’s hardware and software (according to actual situation), the templates are permanently deleted and not remained in any format.

3. Use of Personal Data

3.1. Not being used for a new purpose without prescribed consent

If data user wishes to apply the collected biometric templates to any purpose other than any use authorized, they are responsible for reintroducing to the users and obtain their consents before doing so.

4. Security of Personal Data

4.1. Practicable steps being taken to ensure no unauthorized or accidental access, processing, erasure, loss, use and transfer

ZKTeco’s applied biometric algorithms do not collect complete biometric features of the collected persons, instead only 10 to 50 feature points are collected for calculation. Even if biometric templates are illegally obtained, it is not possible to reorganize them into complete biometric features. In ZKTeco’s software and hardware, various advanced electronic encryption techniques have been applied for the greatest extent of protection to data security.

5. Openness – Information be Generally Available

5.1. Data User should provide policies and practices in relation to personal data.

5.2. Data User should publicize the kinds of biometric data held.

5.3. Data User should inform the main purposes for which biometric data are used.

6. Access to Personal Data

6.1. Data user should know that, all data subjects have the rights to access and correct his or her personal data.

Frequently ask question of biometric data:

I. What is Biometric Data?

Physiological data born with an individual

• DNA samples, fingerprint, palm veins, iris, retina
• Facial images and hand geometries

Behavioral data developed by an individual

• hand writing pattern, typing rhythm, gait, voice

II. Is Biometric Data Personal Data?

Biometric data alone is not regarded as personal data according to law, as it may not reveal identities. However,
if biometric data is stored in a database that links customers/staff members, they are personal data.

TFT Color Screen:

• Audio-visual indication for acceptance and rejection of valid/invalid fingerprints.
• More simple and convenient operation.

Fast and Accurate Fingerprint Algorithm:

• 1 touch a-second user recognition.
• Store 3,000 fingerprint templates, 30,000 cards and 100,000 transactions.

Easy Installation and Connectivity:

• Wiegand input & output.
• Network interface by TCP/IP or RS485

Easy Operation and Management:

• Work without computer.
• Buzzer, 4×4 keypads and door bell.
• Built-in serial and Ethernet ports.
• Built-in USB port allows for manual data transfer

Full Access Control Features:

• Anti-passback. Webserver. Access control interface for 3rd party electric lock, door sensor, exit button, alarm and doorbell

Slim and Elegant Design:

• Sleek and slim design for narrow spaces.

Conclution

Secure identity is a superior requirement of every industrial and construction site, educational institutes, etc. The identity will ensure that your workplace is safe and secured. When identity is confirmed, any workplace will be secured for us for sure.

So, don’t be late to install a fingerprint security system in your working area. Hopefully, this article gives you a clear understanding of the ZKTeco F18 price in Bangladesh and the benefits of this system. Security should get the utmost preference. Install this device, ensure your valuable belongings and other employee’s security.