Basic Info.
Application
Community, Garage, Company, Hotel, Warehouse, Parking, Hospital, Campus, Supermarket, Enterprise, Factory, Park
Image Sensor
1/1.8inch Star CMOS
Image Resolution
1920(H) X 1080(V)
Minimum Illumination
0.001lux(Color), 0.0001lux(Mono)
Automatic Aperture
DC-Iris
Signal to Noise Ratio
More Than 56dB
Wide Dynamic
Support Automatic Mode / Highlight Suppression / B
Image Settings
Saturation, Brightness, Contrast, Sharpness, Image
Transport Package
Carton Box
Product Description
Dynamic Face Recognition System for Bank VIP Management ECF106 Background
With the development of biometric technology, identity authentication technology has gradually changed from traditional card and card authentication to biometric authentication. As a safe, flexible and convenient technical solution, dynamic face recognition and authentication has been applied to more and more scenarios, such as bank counter ID card verification, office building face brushing access control, hotel lobby intelligent welcome, conference and Exhibition face brushing sign in, park visitor face brushing registration, and security control of public places, traffic fortresses and park communities.
Product overview
Dynamic face recognition system is a software and hardware integrated solution. The video stream is collected by the network camera, and the face information in the video stream is detected in real time by using the face algorithm based on the deep learning training, and compared with the face information in the pre created personnel database to quickly and accurately identify the identity information of the target personnel.
This system realizes the functions of standard face collection, face recognition front-end display, personnel database management, recognition record management, equipment management, etc., and can be deployed and used directly. At the same time, it provides rich expansion interfaces, which can flexibly expand business functions through secondary development or interface with third-party system data to meet the
needs of various business scenarios.
System composition
The system is composed of a face recognition module, a front-end display module, a face recognition core service, and a visual management background.
Face recognition moduleIt is usually a network camera or a snapshot machine, which is used to collect the video stream in real time and transmit the video stream or the face snapshot to the face recognition core service. Front-end display moduleDisplay real-time video stream and face recognition results through large screen display. Face recognition core servicesReceive face images or video streams, perform image / video preprocessing, face detection, feature extraction, and face comparison with face database data, and return recognition results. Visual management backgroundIt is used to manage personnel information base, view or export face recognition records, manage equipment configuration and business parameter configuration, etc. System characteristics• Wide recognition distance0.5m ~ 25m, suitable for more diversified application scenarios and meeting various needs• Fast recognition speedThe detection performance of the Ministry of public security is superior, and the single core and single thread face database ratio is up to 10 million levels per second• Pass rate of 100000 points > 98Based on deep learning technology and lifelong learning strategy, the accuracy rate is maintained at a high level.• Combination of dual features makes it more awesomeCreatively combine the classical structural features with the depth features of the neural network, and reduce the error rate of ten thousand points by 25%.• Near infrared & visible lightIt has both near-infrared and visible light recognition algorithms, with stronger light adaptability, and can meet strong light and weak light for different light conditions.• No cooperation requiredBased on deep neural network learning algorithm, it can collect and recognize face information within 30 degrees of face deflection, without deliberate cooperation.• Rich interfaces and easy integrationRich expansion interface, convenient for rapid integration development and business data access.• Simple and flexible deploymentThe deployment package is highly integrated. It can be completed by copying files and simple configuration. You can choose single machine or cloud deployment. Application Scenario
The system can be directly deployed and used, and provides rich data interfaces. Through secondary development, it can realize more abundant business functions, which is applicable to the following scenarios:
Access control system
Traditional access control often requires swiping cards, fingerprints or cooperative face recognition at the entrance and exit to complete identity verification. The dynamic face recognition system can complete face collection and identity authentication during the user's walking, and control the automatic door opening of the access control, so as to bring a more convenient and smooth door opening experience for the user.
Business welcome
When important customers visit, the system can collect customer images in real time and identify them, immediately confirm the customer's identity, display the welcome screen on the welcome screen, and notify the customer manager to come to receive them, so as to bring excellent visiting experience to customers and improve the corporate image. Security, property monitoring, blacklist warning The system supports tracking and recognition of up to 108 faces on the same screen, which is applicable to security and property monitoring. Enter the blacklist face pictures into the system in advance. When the blacklist personnel enter the shooting screen, they can immediately identify and notify the security personnel. In addition, early warning settings can also be set for strangers as required for security monitoring of communities and parks.
High end conference sign inHigh end meetings are usually oriented to important guests, with high requirements for venue layout and reception details. Signing in is an important initial step. This system allows guests to complete the check-in without signing, code scanning or specific action cooperation, accompanied by the large screen welcome screen, which greatly improves the user experience.
Internal time attendance of enterprisesTraditional time attendance machines require employees to punch in by swiping their cards, fingerprints or cooperative face recognition. However, in rush hours, there is a large flow of punch in people, and queuing often occurs. The system can recognize multiple faces in the screen at the same time, and multiple employees can complete the attendance punch in an instant when passing by at the same time, which is fast and convenient.
Customer source analysis of supermarkets and retail storesCustomer information mining is very important for retail stores. With the help of this system, a database can be established for the members of the commercial supermarket. By capturing the face images of customers in the store in real time, member identification, passenger flow statistics, customer group attribute analysis, passenger flow movement and average dwell time analysis can be carried out, providing data basis for the optimization of the marketing plan of the commercial supermarket.
Product Functions
The main functions of the system are as follows:
| Serial number | Function | Description |
| 1 | Face recognition client display | Display video stream and recognition results |
| 2 | Personnel information management | Manage personnel information, mainly facial features, and support the creation of multiple personnel databases |
| 3 | Face recognition | Perform face detection and feature extraction on the real-time collected face images, and communicate with people
Face to face ratio is correct, and the recognition result is obtained |
| 4 | Identification record | Including records of successful identification and records of failed identification. Detailed record identification
Room, snapshot, template, personnel name and other information |
| 5 | statistical analysis | Statistics: personnel base overview, equipment overview, number of identification, trend of human flow, and number of people
Regional distribution of staff and other data |
| 6 | user management | Set the account, role, authority and other information of login management background |
| 7 | device management | Manage acquisition terminal (network camera) and server equipment |
| 8 | Parameter management | Set face recognition threshold, face deduplication time and whether the client displays strange
Human parameters |
| 9 | System maintenance | Comprehensively monitor the operation of the system and keep various operation logs |
| 10 | Extension interface | Rich expansion interface, convenient for secondary development of business functions or docking with third-party systems
System to meet business needs |
In order to meet the needs of meeting guests in the hall, real-time monitoring and other scenes, the system provides a face recognition client display function, mainly including the display of video streams and recognition results, as well as simple and high-frequency interactive operations. Video stream
Display the real-time video stream from the network camera. If multiple cameras are deployed in the system, the video source to be displayed can be selected by yourself. Identification results
When the system identifies the personnel in the personnel library, the personnel photos and names will be displayed through the interface pop-up window. At the same time, the page will display the list of the most recently identified personnel and the identification time are displayed.
Personnel information managementPersonnel base
The person database is used to manage the basic information of the person to be recognized, especially the face information, so as to compare the real-time collected face with the face database.
Users often perform different business actions for different types of personnel, such as making different welcome speeches for VIP and ordinary users; For example, personnel of different organizations are allowed to enter different access control. In order to facilitate management, the system supports the creation of multiple personnel libraries. A person can belong to multiple personnel libraries at the same time. Personnel information field
The personnel database can store various basic information of personnel, including personnel name, face picture, business number, etc. The personnel information field supports secondary development and extension to meet various business requirements in actual application scenarios. Personnel addition
Point to add personnel information to the personnel library. Single person or batch addition is supported.
Individual personnel can be added directly through the software interface; Batch addition requires editing and uploading personnel information according to the specified template. In addition, single or batch personnel addition through standard interfaces is supported. Personnel deletion
It refers to deleting the personnel information in the personnel library. Individual deletion or batch deletion of personnel is supported. Personnel information modification
It supports editing various information of each person. Personnel information query
Support personnel search by ID, name, etc., or filtering by personnel database. Personnel information synchronization
The system provides a personnel information synchronization interface to support the docking of business data with the system data during secondary development.
Face recognition
The system uses the network camera as the information source of face recognition, and sends the real-time collected pictures or videos to the face recognition system. The system automatically completes face detection, feature extraction and face comparison, and obtains the recognition result according to the comparison score. Parameter setting
Support the setting of technical parameters related to face recognition, such as the storage address of face images, the number of face search returns, and the face comparison threshold. Business expansion
According to business needs, specific business actions can be secondary developed through the interface, such as display of identification results, VIP welcome, blacklist alarm, access control and door opening, and attendance clocking.
Identification record
Face recognition records include records of successful recognition and records of failed recognition. Recognition success refers to matching the face in the snapshot to the person in the personnel database; recognition failure refers to detecting the face from the snapshot but not matching the person in the personnel database.Identification record information
The record information of each successful identification includes: identification time, personnel name, captured picture, template picture, comparison score, equipment location, etc.
The record information of each identification failure includes: identification time, captured pictures and equipment location. Identification record query
Support identification record search by ID and name, or filter by time period, camera and identification result.
Statistical analysis
Statistical analysis includes statistical overview and face recognition data statistics.
The statistical dimensions mainly include: personnel base overview, equipment overview, number of identified persons, trend of human flow, regional distribution and type distribution of identified persons.
Through appropriate combination filtering, you can view the trend of pedestrian flow in specific / all stores, the comparison of pedestrian flow in different stores, and the proportion of different customer types in a certain period of time. The age distribution and gender distribution of customers will also be added in subsequent versions.
At present, the data related to face recognition is only limited to the number of successful recognition / person times, and does not include strangers temporarily. The statistics of strangers will be gradually increased in subsequent versions. Overview of personnel base
Displays the total number of people in the personnel pool, the number of personnel pool groups, the number and proportion of people in each personnel pool group, and whether the personnel information is complete. Equipment Overview
Display the total number of cameras, online devices and offline devices connected to the system. Identification number
Display the number of people identified yesterday and today and the number of people identified.Trend of passenger flow
Display face recognition data within a certain period of time. It supports filtering of identification equipment, personnel base and start / end time. You can select to view the number of identification personnel or the number of identification personnel. Regional distribution of identified personnel
Display the comparison of the number of people identified by all cameras connected to the system. Since each camera represents a corresponding branch or area, this module can be used to check the number of people identified in each area. Identify the type distribution of personnel
It displays the comparison of the number of personnel in each personnel pool among the identified personnel in the system.
User management Account management
Account refers to the user account that logs in to the management background of the system, usually the administrator. Account information includes: login account, name, role and mobile phone number.
It supports setting roles for accounts and configuring different management permissions for roles to facilitate batch management of account permissions. Role management
The role name is defined by the administrator, and the permissions are flexibly configured as required. Each account can be set with multiple roles, and each role can be associated with multiple accounts. You can add, delete, modify, and enable / disable roles.
It supports searching roles by name, permission and character, and filtering roles by time period and status. Permission management
In order to facilitate administrators to manage business flexibly, the system defines multiple permissions according to different businesses. Each permission is specific to a specific business. You can configure a series of permission combinations for administrators as required.
When logging in to the system, users can only view the relevant business pages for which they have permission, and other pages are not displayed. Device management
Equipment management mainly includes camera management and host management.
Camera management
The system will record the information of all cameras currently connected, including device name, device type, whether online, device IP, installation location, RTSP address, etc.
Supports adding, deleting devices and editing device information.
Support detailed device parameter configuration, such as audio and video settings, time settings, network settings, serial port settings, push settings, data query, device maintenance, etc.
The equipment management of the system is mainly business parameter management. Click "login equipment" to enter the camera independent configuration platform for network, audio and video and other general equipment function configuration. Host management
The host management function is not included in the current version, and will be added in later iterations.
Parameter managementFace recognition thresholdUsed to set the comparison threshold between the face snapshot and the base photo. The higher the threshold, the lower the error rate; The lower the threshold is, the less recognition the lower the rateFace deduplication time
When a person continuously or repeatedly appears in the camera screen, multiple recognition records may appear. In order to avoid multiple records of the same person in a short time, the face removal time can be set. Within a certain time, the same person will not be recognized repeatedly. The front-end interface display stranger identification records or not
This system provides the front-end display function of face recognition, mainly including the display of video streams and recognition records. Through this parameter, you can set whether the recognition record only includes the successful person or also includes strangers. - 1000 Face registration and warehousing shall not exceed 10 minutes recognition speed
- 10000 level face database 1: n comparison speed within 1 second1: N interface comparison speed (within 1s)
- 1: (1) interface comparison speed: video stream loading speed within 1s
- Real time video streaming client shows the reliability within 1 second of delay
- The system operates continuously for 7 * 24 hours deployment environment
- CentOS 7.6.1810
- Ubuntu 16.04 or above
- OpenSUSE 15
- Linux RedHat Running Environment
- Face recognition front end: Windows / Linux platform client
Background Management: multi Platform browser face Image specification- Single face image is not less than 30K and not more than 1 Megabyte
The distance between two eyes shall not be less than 20 pixels (system capacity)- Support massive user registration (depending on specific application scenarios and storage space)
Hardware equipmentThe hardware equipment recommended for the system is as follows: Face capture camera
| Type | Item | Parameter |
Image | Image sensor | 1/1.8Inch star CMOS |
| Image resolution | 1920(H) x 1080(V) |
| Frame rate | 1 ~ 25FPS, default 25FPS |
| Minimum illumination | 0.001Lux(Color), 0.0001Lux(Mono) |
| Day night switching | IR-cut automatic switching, manual switching, timing switching, alarm input switching |
| | Lens interface | C/CS |
| Automatic aperture | DC-IRIS |
| Dynamic range | 120dB |
| Signal to noise ratio | >56dB |
| White balance | Automatic |
| Wide dynamic | Support automatic mode / highlight suppression / backlight compensation |
| Image settings | Saturation, brightness, contrast, sharpness, image flip can be adjusted by browser |
| OSD stack | Text title, time, etc |
| Code | Video compression standard | H.265/H.264/MJPEG |
| Video resolution | Main code stream: 1920×1080/1280×720/640×360/352×two hundred and eighty-eight |
| Subcode stream: 720×576/640×360/352×two hundred and eighty-eight |
| Video bit rate | 512Kbps ~ 10000Kbps |
| Network | Network protocol | HTTP/TCP/ARP/RTSP/RTP/UDP/SMTP/FTP/DHCP/DNS/PPPOE |
| Image push | Support automatic push of HTTP and FTP protocols |
| Interface protocol | GB/T28181, ONVIF |
Interface | Network interface | 1Gigabit RJ45 Ethernet interface |
| Serial port | 1Way RS485 |
| Storage interface | 1Way TF card socket |
| Alarm input | 2Channel active input |
| Alarm output | 2Two active outputs and two passive outputs |
| Video interface | 1Channel CVBS analog video output |
| Reset button | support |
| Audio interface | 1One 3.5mm standard audio input (line in) and one 3.5mm standard audio output (line
Out), 1 built-in microphone |
Intelligence | Face area | Detection area setting, shielding area setting |
| Face capture | Minimum pupil distance configuration, maximum pupil distance configuration, capture times / capture interval configuration |
| Face optimization exposure | Support |
| Face tracking | Support 108 faces on the same screen |
| Face capture | Support face scoring, frame by frame tracking, automatic screening of the best face, support horizontal direction-
90 ° ~ + 90 ° and - 45 ° ~ + 45 ° in the vertical direction, |
Face data structure
turn | Support, including face occlusion, gender, age, etc |
| Human shape detection | Support, including back body detection, moving human figure discovery without face (face completely occluded) |
| Intelligent alarm | Cross boundary detection, area intrusion detection, scene change detection, face detection, virtual focus detection, object leftover detection, object pickup detection, illegal parking detection, personnel gathering detection, retrograde detection, wandering detection,
Fast movement detection, entering area detection, leaving area detection |
| Reliability | Power supply | DC:12V±20% ;AC:24V±20% ;PoE:802.3af |
| Power waste | 5W (excluding fill lamp) |
| Working temperature | -20°C~+70°C |
| Type | Item | Parameter |
| Model | YSCXD900-J1 | See the following table for details |
Hardware | operating system | linux |
| CPU | Intel Core i7-8550u |
| Memory | DDR3L 8G |
| Hard disk | M. 2 solid state 256g |
| Interface | Built in dual band wireless WiFi (2.4G + 5g), Bluetooth, usb30*2 USB2. 0 * 2, HDMI interface |
| Size | 12.9*12.5*3.6cm |
Software | Video analysis | Support rtps protocol analysis of video stream |
| Face detection | Support face image detection and location |
| Face tracking | Support tracking and positioning in the process of face movement |
| Face clipping | Support cutting out partial photos of face images from video frame images |
| Feature extraction | Support feature point extraction of face image |
| Face comparison | Support the comparison function of face feature database |
| Log query | Support log query function of face recognition process |
Address:
8f, Building1, Huihuang International Plaza, Shangdi 10th Street, Haidian District, Beijing, China
Business Type:
Manufacturer/Factory, Trading Company, Group Corporation
Business Range:
Computer Products, Security & Protection
Management System Certification:
ISO 9001, ISO 14001
Company Introduction:
Founded in 1997, Eyecool Technology (formerly Techshino) focuses on the R & D, application and service of multi-modal biometric technology and products. It owns in-house developed single algorithms ranging from fingerprint, face, iris, finger vein to multi-modal biometric algorithm including face and iris fusion, fingerprint and finger vein fusion. Eyecool Technology is committed to provide enhanced digital identity authentication and management for high-security, high-privacy and big database scenario application.
With 25+ development, Eyecool Technology accumulated more than 400 patents and participated in the formulation and revision of 82 national/industry biometric application standards, and the core technology won the second place of the National Science and Technology Invention Award.
The fingerprint recognition algorithm and scanners were selected into the recommended list of application algorithms and equipment by the Ministry of Public Security for resident identity card related application, in domestic market. Fingerprint algorithm passed the compatibility test of NIST MINEXIII. Related fingerprint scanner passed PIV, FIPS 201 and STQC. The face recognition algorithm of Eyecool won the second place in LFW, an international face algorithm evaluation competition. And the latest iris recognition algorithm ranked in 3rd place in the NIST IREX 10, an international iris algorithm evaluation hold by National Institute of Standards and Technology in America. It also won the second place in the international iris recognition algorithm open competition (NICE: II) in 2009. A variety of iris scanners are STQC-certificated.
Eyecool Technology provides a whole set of biometric technology, products and services ranging from algorithm, SDK, an unified platform, application software, to intelligent terminals. Until now, it has already serves more than 2000 customers worldwide, covering finance, public security, government, military, education, civil aviation, enterprises and institutions etc. In financial sector, Eyecool Technology delivered its products and service to more than 150 banks across the whole country with 180, 000 outlets and its market share reaches 80% in biometric application. In addition, Eyecool Technology′s products are widely used in more than 100 colleges and universities, 20+ provinces and cities public security.
Company History
In 1997, Company was founded.
In 1998, stared R&D of algorithm.
In 2007, succeeded in algorithm R&D, including fingerprint, face and iris; Meanwhile first fingerprint application was deployed in head office in PSBC.
In 2012, Company won 2ND place in National Science and Technology Award and Online Face Recognition application in national ID management was implemented in bank.
In 2015, face recognition system and product were applied in 20+ banks in a large-scale
In 2016, multi-modal biometric identification system was firstly used in Bank of Communications; Besides, Eyecool became one of recommended suppliers for Police National ID fingerprint capture and fingerprint algorithm. In the same year, Company changed name from Techshino to Eyecool
In 2018, Eyecool group located its HQ in Xiong′an and its business expanded from finance to police, army, campus, smart City etc.
Until now, Company takes 80% market share in banks in China; And products was sold to Columbia, Indonesia, India, Philippines etc. More than iris 10, 000 devices sold abroad.
){kind=link}