ARMATURA AMT-FAPVR-21 Built-in High Performance Storage User Manual

User Manual
AMT-FAPVR-21
Date: October 2022
Doc Version: 1 .2

AMT-FAPVR-21 Built-in High Performance Storage

Thank you for choosing our product. Please read the instructions carefully before operation. Follow these instructions to ensure that the product is functioning properly. The images shown in this manual are for illustrative purposes only.
For further details, please visit our Company’s website www.armatura.us.

Copyright © 2022 ARMATURA LLC. All rights reserved.
Without the prior written consent of ARMATURA LLC no portion of this manual can be copied or forwarded in any way or form. All parts of this manual belong to ARMATURA and its subsidiaries (hereinafter the “Company” or “ARMATURA”).

Trademark
is a registered trademark of ARMATURA LLC. Other trademarks involved in this manual are owned by their respective owners.

Disclaimer
This manual contains information on the operation and maintenance of the ARMATURA product. The copyright in all the documents, drawings, etc. in relation to the ARMATURA supplied product vests in and is the property of ARMATURA. The contents hereof should not be used or shared by the receiver with any third party without express written permission of ARMATURA.
The contents of this manual must be read as a whole before starting the operation and maintenance of the supplied product. If any of the content(s) of the manual seems unclear or incomplete, please contact ARMATURA before starting the operation and maintenance of the said product.
It is an essential pre-requisite for the satisfactory operation and maintenance that the operating and maintenance personnel are fully familiar with the design and that the said personnel have received thorough training in operating and maintaining the machine/unit/product. It is further essential for the safe operation of the machine/unit/product that personnel have read, understood, and followed the safety instructions contained in the manual.
In case of any conflict between terms and conditions of this manual and the contract specifications, drawings,
instruction sheets or any other contract-related documents, the contract conditions/documents shall prevail. The contract specific conditions/documents shall apply in priority. ARMATURA offers no warranty, guarantee, or representation regarding the completeness of any information contained in this manual or any of the amendments made thereto. ARMATURA does not extend the warranty of any kind, including, without limitation, any warranty of design, merchantability, or fitness for a particular purpose.

ARMATURA does not assume responsibility for any errors or omissions in the information or documents which are referenced by or linked to this manual. The entire risk as to the results and performance obtained from using the information is assumed by the user.
ARMATURA in no event shall be liable to the user or any third party for any incidental, consequential, indirect, special, or exemplary damages, including, without limitation, loss of business, loss of profits, business interruption, loss of business information or any pecuniary loss, arising out of, in connection with, or relating to the use of the information contained in or referenced by this manual, even if ARMATURA has been advised of the possibility of such damages.
This manual and the information contained therein may include technical, other inaccuracies, or typographical errors. ARMATURA periodically changes the information herein which will be incorporated into new additions/amendments to the manual. ARMATURA reserves the right to add, delete, amend, or modify the information contained in the manual from time to time in the form of
circulars, letters, notes, etc. for better operation and safety of the machine/unit/product. The said additions or amendments are meant for improvement /better operations of the machine/unit/product and such amendments shall not give any right to claim any compensation or damages under any circumstances.
ARMATURA shall in no way be responsible (i) in case the machine/unit/product malfunctions due to any non-compliance of the instructions contained in this manual (ii) in case of operation of the machine/unit/product beyond the rate limits (iii) in case of operation of the machine and product in conditions different from the prescribed conditions of the manual. The product will be updated from time to time without prior notice. The latest operation procedures and relevant documents are available on http://www.armatura.us.

If there is any issue related to the product, please contact us.

ARMATURA Headquarters
Address
190 Bluegrass Valley Pkwy,Alpharetta, GA 30005. USA
For business-related queries, please write to us at info@armatura.us.
To know more about our global branches, visit www.armatura.us.

About the Company
ARMATURA is a leading global developer and supplier of biometric solutions which incorporate the latest advancements in biometric hardware design, algorithm research & software development.
ARMATURA holds numerous patents in the field of biometric recognition technologies. Its products are primarily used in business applications which require highly secure, accurate and fast user identification.
ARMATURA biometric hardware and software are incorporated into the product designs of some of the world’s leading suppliers of workforce management (WFM) terminals, Point-of-Sale (PoS) terminals, intercoms, electronic safes, metal key lockers, dangerous machinery, and many other products which heavily rely on correctly verifying & authenticating user’s identity.
About the Manual
This manual introduces the operation of AMT-FAPVR-21.
All figures displayed are for illustration purposes only. Figures in this manual may not be exactly consistent with the actual products.

Document Conventions

Conventions used in this manual are listed below: GUI Conventions

For Software

Convention| Description
Bold font| Used to identify software interface names e.g. OK, Confirm, Cancel.

| Multi-level menus are separated by these brackets. For example, File > Create > Folder.
For Device
Convention| Description

< >| Button or key names for devices. For example, press . [ ]| Window names, menu items, data table, and field names are inside square brackets. For example, pop up the [New User] window. /| Multi-level menus are separated by forwarding slashes. For example, [File/Create/Folder].

Symbols

cautionary example.

Product Introduction

1.1 Overview
AMT-FAPVR-21 is a dual-camera multimodal standalone module for face and palm recognition empowered by high-performance processors. The multimodal standalone module includes functions such as image collection, biometric template extraction, matching, transmission, and data management. The module has compact, light-weight size and is highly adaptable to various ambient lighting and supports touchless multi-biometric recognition.
The dual-camera module is equipped with proprietary ARMATURA facial and palm recognition algorithms and supports both facial and palm recognition in different application scenarios. The facial recognition algorithm supports face detection, liveness detection, face mask detection, facial recognition, age recognition, gender recognition, and facial expression recognition. The palm recognition algorithm supports palm detection, live palm detection, and palm recognition.
The dual-camer  module integrates near-infrared light and support dual-channel image collection with visible and near-infrared light. Both lenses use low light and wide dynamic range image sensors. They are excellent in adapting to ambient light with the help of powerful image processing algorithms. A visible-light lens can capture high quality face images in low illumination (0.01 lux) and bright light (≤ 50,000 lux) as well. A near-infrared lens can capture high-resolution, highcontrast, and distortion-free palm lines
and subcutaneous vein images, considerably improving the recognition accuracy.
The dual-camera module has built-in face and palm algorithms, flexible architecture, and are suitable for various scenarios. The module provides three application modes for deployment in different scenarios. They are:

1. The USB video class (UVC) functions as a collection module to stream video.
2. The module does the video streaming and face/palm biometric extraction, while the host does the biometric comparison.
3. After completion of video streaming, extraction, and biometric comparison in the module. Select a deployment mode based on application requirements and platform features to make the best use of dual-camera collection.

The software development interfaces allow quick integration of face and palm recognition functions to an application platform. The functions apply to areas such as attendance checking, entrance control, and channel access.

1.2 Features

• Dual low light and wide dynamic range image sensors, maximum 105 dB dynamic range.
• Support Face/Palm AE (smart brightness enhancement technologies for facial and palm recognition).
• Built-in near-infrared light supports smart switch control based on image detection algorithms.
• Captures face and palm images at 25 frames per second
• Maximum face and palm capacities are 100,000 and 10,000.
• Wide field-of-view allows its cameras to recognize individuals at wide varying heights.
• Include SDKs for major platforms (Windows, Android, and Linux).
• Support UVC (USB video device) streaming video protocol and Human Interface Device (HID) data protocol over USB 2.0.
• Compact, light-weight size with USB 2.0 interface makes for easy integration with a wide range of host hardware devices
• Provide a touchless, hygienic, and non-invasive biometric identification solution.
• Flexible and open software architecture deployment, adaptable to various application scenarios.
• Integrated communication and power supply USB interface.

Product Specifications

2.1 Technical Specifications

infrared light)
Communication Interface| USB 2.0
Communications Protocol| Composite UVC device + HID protocol
Camera Type| M8, 2MP
Required Power Supply| DC 5V/1.1A
**** Collector Size| 80 mm x 44.4 mm x 29.78 mm (Length x Width x Height)

2.2 Electrical Features

Specifications| Test Conditions| Min| Standard| Max| Unit
---|---|---|---|---|---
Operating Voltage| –| 4.75| 5.0| 5.25| V
Operating Current| T = 25℃, VCC = 5.0 V|

–

| ****

–

| ****

1.1

| ****

A

Operating Power Consumption| T = 25℃, VCC = 5.0 V| ****

–

| ****

–

| ****

5.5

| ****

W

Standby Power| T = 25℃, VCC = 5.0 V|

–

| ****

–

| ****

0.6

| ****

A

Standby Power Consumption| T = 25℃, VCC = 5.0 V| ****

–

| ****

–

| ****

3

| ****

W

Operating Temperature| –| –10| –| 50| ℃
Storage Temperature| –| –40| –| 85| ℃

2.3 Optical and Image Specifications

480 x 640@25fps
Default Output Format| MJPEG| NV12
Signal-to-Noise Ratio| 50 dB| 50 dB
Dynamic Range| 105 dB| 105 dB
Min Illumination| 0.01 lux| 0.01 lux
Max Illumination| 50,000 lux| 20,000 lux
Image Distortion Rate| ≤ 0.5%| ≤ 0.5%
Aperture| F2.0| F2.0
Vertical Viewing Angle| |
---|---|---
Horizontal Viewing Angle| |

2.4 Model Specifications

Features
Supported Algorithm Type| Facial Recognition| Palm Recognition
Effective Image Range
(Algorithm Recognition
Range)| 50cm – 160cm| 15cm – 40cm

Algorithm Specifications

3.1 Facial Recognition Algorithm
Based on years of in-depth technological development in the biometrics industry, ARMATURA’s facial recognition algorithm supports precise recognition of multiple attributes of all age groups in complex light conditions. It can detect up to 180 facial key points within milliseconds, such as eyes, lip edges, nose tips, and contours. The facial recognition algorithm supports face detection, liveness detection, face mask detection, age recognition, gender recognition, facial expression recognition, and facial recognition.
The algorithm can detect the features on a variety of postures. As shown in Figure 3.1 below: 3.1.1 Face Detection
The algorithm can detect a human face within a range of 2m with a dynamic speed. It helps in fast and touchless verification.

3.1.2 Fake Face Detection
The algorithm can distinguish whether the detected face is a real alive face or a mask or cover concealing the real identity. It helps counter spoofing.

Features| Dual-lens liveness detection (maximum range: 1.6 m)
Visible-light lens detection (maximum range: 2 m)
---|---
Detection Speed| 30 ms
Min. Image Size| 128 x 128 pixels
Posture Adaptability| Yaw ≤ 30°, Pitch ≤ 30°, Roll ≤ 45°
Precision| TAR=99.2% when FAR=0.001%

3.1.3 Mask Detection
The algorithm is capable of detecting if a person has worn a mask or not. It helps identify such person and take required action.

3.1.4 Facial Attribute Recognition
The algorithm is also capable of understanding facial attributes such as age, gender, and facial expression. It can display all the above-mentioned attributes when it detects a face, on the go.

Disgusted
Maximum Range| 2 m
Detection Speed| 20 ms
Min. Image Size| 128 x 128 pixels
Posture Adaptability| Yaw ≤ 30°, Pitch ≤ 30°, Roll ≤ 45°
Precision| Age recognition (±5 years) > 85%, Gender recognition > 96%, Facial expression recognition > 88%

3.1.5 Facial Recognition
The algorithm can detect a face and recognise the identity of the person within fractions of second and with high accuracy rate.

3.1.6 Facial Registration Image Quality Requirements
The image saved while face registration has JPG format and the minimum resolution is 128 x 128 pixels. The individual should take care of the following things while registering their face:

• They should not wear a face mask or colored glasses.
• They should ensure that the frame of the glass does not block the eyes and that the lenses are non-reflective.
• The face angle tolerance for pitch, roll, and yaw should be between ±10 degrees.
• The whole face should be in the camera frame and the aspect ratio of the face must not be distorted.
• Face contours should be clear and the light must be even.
• The facial expression should be natural without obvious exaggeration.

3.2 Palm Recognition Algorithm
Palm recognition technology uses near-infrared light reflections to obtain palm line and subcutaneous vein biometrics and utilizes convergence to identify or verify individual identities. The technology boasts powerful anti- spoofing functionality, as only flowing  blood can form vein images.
AMT PalmLite 12.0 is an excellent near-infrared palm recognition algorithm developed for largecapacity palm recognition with a high tolerance of palm postures in complex light conditions. The algorithm focuses on improving adaptability to an extensive variety
of usage environments and user habits and ensuring high accuracy. In this manner, the robustness and pass rate are considerably improved.
The algorithm applies to various palm postures as shown in Figure 3.3: 3.2.1 Palm Recognition Specifications

3.2.2 Palm Registration Image Quality Requirements
The image saved while palm registration has JPG or BMP format and the minimum resolution captured from the device is 480 x 640 pixels. The individual should take care of the following things while registering their palm:

• The palm and background must be separated.
• The light must be even so that the veins must be visible.
• The whole palm (including the wrist) should be in the camera frame and the aspect ratio of the palm must not be distorted.
• The palm angle tolerance for pitch, roll, yaw, and bend should be ±5 degrees.

Application Scenarios

The AMT-FAPVR-21 dual-lens camera collectors are designed with a complete metal structure shell and corresponding fixed brackets, facilitating quick installation and extension on an existing device, or integration into a device. Based on the integrated design, the dual-lens camera collector can be connected to an application platform device using its own dual-port USB-TYPE-A cable. In this way, the connection between the power supply and communications device is accomplished which simplifies the design considerably. The example of an application system is shown in Figure 4.1 below:

To improve platform adaptability and to lower development difficulty, the dual-lens camera collectors integrate with all the functions in the biometric recognition application process, including image collection, image transmission, template extraction, template comparison, template management, and comparison result output. All the functions can be configured using the software. You can develop a required application program based on the SDK to achieve your desired functionality and complete application design.

• Image Collection and Transmission
  The collection and transmission of the face and palm images are made by a collector. You can specify different resolutions for image output. For a list of supported resolutions and frame rates, see section 2.3 Optical and Image Specifications. Image transmission adopts the standard UVC protocol. Visible and near-infrared light images are transmitted using separate UVC ports. You can choose the ports as per with image requirements.

• Biometric Template Extraction Biometric template extraction is the most tedious biometric recognition process. To reduce the impact of the collectors on platform performance and improve the collector’s platform adaptability, the biometric template extraction is performed in the collector by default. The high-performance
  heterogeneous processor of the collectors helps free up the computing resources of your platform.

• Template Comparison and Management Comparison and management of biometric templates can be performed in the collectors or on a platform, according to your requirements.

The collectors provide three application modes for use in different scenarios:

1. The UVC functions as a collector to stream video.
2. The video streaming and face/palm biometric extraction is achieved in the collector, while the biometric comparison is performed by a host.
3. The video streaming, extraction, and biometric comparison are achieved in the collector.

If you use the platform for template comparison and management, the collector will push the template to the platform using the Human Interface Devices (HID) port after a biometric template is extracted.
This applies to scenarios in which template storage is strictly restricted. Figure 4.2 shows some functions of a back-end comparison system. If you use the collector for template comparison and management, the collector will only send the output of the comparison result through the HID port for a client application to call. This can minimize the computing load of the platform processor. As such, it is particularly suitable for the integration of the face/palm recognition function in low-performance embedded platforms. Figure 4.3 shows some functions of a front-end comparison system. As the SDKs are provided, you only need to call the function interface of a configuration and write an application to integrate face/palm recognition function in your platform. It shortens the development cycle and improves product competitiveness.

Structural Dimensions

Cable Requirements

The dual-lens camera collector is equipped with wires by default. For special needs, customers can customize according to the following requirements:
The total length of the USB cable is 1.2 meters and has a dual USB Type-A power supply interface design. A shielding layer is required, and the number of weaves should be 64, and a D+/D- twisted pair of 26 AWG cables. The wire has a magnetic ring on the collector side. Figure 6.1 shows a design example for your reference. ![ARMATURA AMT-FAPVR-21 Built-in High Performance Storage

• Cable Requirements](https://manuals.plus/wp-content/uploads/2023/12 /ARMATURA-AMT-FAPVR-21-Built-in-High-Performance-Storage-Cable- Requirements.jpg)

Design Guide

7.1 Optical Design
Based on integrated design considerations, the lens surface of the dual-lens camera collector has been covered with a transparent acrylic cover. If there is no special requirement, it is not recommended to add any other cover which may affect the light transmission performance.
7.2 Heat Dissipation Design
The dual-lens camera collector adopts an all-metal shell design and surface heat dissipation stripe so that it has good heat dissipation performance. It is recommended, to use it in an open and ventilated environment to improve the heat dissipation efficiency of the collector and to better exert the performance of the collector.

Installation Guide

8.1 Installation Modes
The dual-lens camera collector supports a variety of installation modes, including front-locking, rearlocking, down-locking, and desktop mode. Users can choose and design flexibly according to their needs. The following is a detailed description of each installation mode.
8.1.1 Front-locking and Rear-locking
The front-locking and rear-locking modes have the same fixing methods, only the installation directions are opposite to each other. The installation and fixing method are shown in Figure 8-1 using the example of rear-locking mode: 8.1.2 Down- locking
The down-locking mode can facilitate customers to directly integrate the dual- lens camera collector on the top of the existing device. It helps in the quick integration of the collector. The specific installation method is shown in Figure 8-2. 8.1.3 Desktop
The desktop installation mode is mainly to meet customer desktop-level extensions and applications. This installation mode needs to be fixed with the corresponding fixing bracket. By adding accessories at the bottom of the dual- lens camera collector, it can be adjusted to the 1/4 screw of the traditional security bracket. The installation method and the overall outcome are shown in Figure 8-3.

8.2 Installation Height and Angle
When you install a dual-lens camera collector, you need to consider the influence of the installation angle and height on the field of view for image collection. If the mixed face/palm recognition algorithm is used, you need to consider the field of view that faces can adapt to, the range requirements of palm recognition, and the convenience and speed of man-machine interaction. Therefore, selecting a proper installation angle and height is crucial to the front-end image collection and user experience of the entire system.
Figure 8.4 describes basic concepts related to installation height and angle. Lens optical axis (a): a line that defines the path along which light propagates through a lens.
Reference plane (h): a plane on which a person stands when a reader is used.
Installation height (H): the height above the reference plane, at which a reader is installed.
Installation angle (b): the angle between the lens optical axis (a) and the reference plane (h), within a range of –90° to 90°.
Recognition distance (S): the distance between a face and a lens when a collector is used.

The purpose of optimizing the installation height (H) and installation angle (b) is to improve the collector’s adaptability to the height of a human body and user experience within the normal range of man-machine interaction. Human body height is an important reference index because it is linearly related to the height of a person’s face and the length of a person’s arms. Therefore, if the installation design of a collector meets the requirements of adaptable body height, it also meets the requirements of facial and palm recognition.
To facilitate installation design, we recommend using a horizontal or slant installation mode.

8.2.1 Horizontal Installation Mode
In horizontal installation mode, the installation angle (b) is set to 0°, and the installation height (H) is set to 1.5 m. If a person can be recognized at a distance of 0.5 m to 0.8 m, the reader can adapt to individuals with a height range between 1.38 m and 2.0 m, as shown in Figure 8.5. 8.2.2 Slant Installation Mode
In the slant installation mode, the installation angle (b) is set to 10°, and the installation height (H) is set to 1.45 m. If a person can be recognized at a distance of 0.5 m to 0.8 m, the reader can adapt to individuals with height is between 1.45 m and 2.18 m, as shown in Figure 8.6.

https://www.armatura.us
190 Bluegrass Valley Pkwy,
Alpharetta, GA 30005, USA
E-mail: info@armatura.us
www.armatura.us
Copyright © 2022 ARMATURA LLC. All Rights Reserved.

References

• ZKTeco | Biometrics and Computer Vision
• Armatura | Global Biometric Technology Provider

Read User Manual Online (PDF format)

Read User Manual Online (PDF format)  >>

Download This Manual (PDF format)

Download this manual  >>