Popular products
speech recognition
Face recognition
Character recognition
Content security
Analysis of emotional tendency
information community file
Console
Offline face recognition SDK
It provides localized face detection and collection, multi-mode live detection, face comparison and recognition and other functions on the device side, and can perform offline authentication, driver status analysis, attention detection, face attribute analysis and other functions in a network free environment
preferential Cloud based optimization, special offers, as low as 59 SDKs, instant rush purchase>> >
Use Now
Technical Documentation
product price
Function introduction
Technical features
Specification information
Application scenarios
Product scheme selection
Selection of living scheme
Product pricing
Related product recommendation
Function introduction
Face detection and tracking
Real time monitoring of the face in the video stream under the offline state of the device side, while supporting the processing of static pictures or video streams, and continuous tracking of the currently detected face, dynamic positioning of the face contour, stable fitting face.
Face Key
Continuously track the currently detected face and dynamically display the core key points on the face in real time, which can be used for facial features positioning, dynamic stickers, video special effects, etc.
Face quality monitoring
In the process of face detection and tracking, verify the pose angle, occlusion, definition and lighting conditions of the face in real time, and only those that meet the quality conditions will be collected.
Face image acquisition
For the video stream, the real-time face image collection is completed, and the output meets the quality filter conditions. You can customize the collection face size, collection frequency, collection quality and other settings.
Offline RGB live detection
For the video stream/picture, we can judge whether the target object is alive by collecting the flaws of the portrait (moire, imaging malformation, etc.), which can effectively prevent cheating attacks such as screen second remake, and can use single or multiple judgment logic.
Offline near-infrared in vivo detection
For the video stream/picture, the near-infrared imaging principle is used to realize the live judgment at night or without natural light. Its imaging characteristics (such as the inability of the screen to image, different reflectivity of different materials, etc.) can achieve highly robust live judgment.
Offline 3D structured light live detection
Judge whether the target object is alive through 3D modeling. Based on the principle of structured light imaging, it can effectively defend against images, videos, screens, molds and other attacks.
Offline comparison and identification
It provides localized 1:1 comparison, 1: N search functions, and highly robust algorithms. It can deal with most of the pose, skin color, lighting and other environments, and can be effectively applied to person ID comparison, identity verification and other scenarios.
Offline face database management
It provides localized face group, user, and face image three dimensional face database management function. There is no limit to the size of the face database. It can be customized to use the performance of visual devices. It is recommended that the number of faces be less than 10000.
Technical features
Rich function integration
It integrates face recognition related detection, alignment, live and recognition models and functions, and supports attribute recognition, driver behavior analysis and attention detection functions
Support multi-mode live detection
It supports live detection and recognition of RGB (visible light), NIR (near infrared) and 3D structured light modes, and can be flexibly combined and configured according to the characteristics of the use scene
Perfect adaptation of mainstream hardware modules
Adapt to the mainstream core scheme and lens module, good compatibility can be applied to a wider range of industry scene needs, greatly shortening the project landing cycle
Specification information
Application scenarios
Certification machine
Face access control machine
Face Attendance Machine
Face payment machine
Certification machine
The device reads ID chip photos and collects face photos on site. First, it judges the face photos collected on site based on the live detection model, and then conducts 1:1 comparison between chip photos and face photos to complete identity verification. This equipment can be used for identity verification in government affairs, customs clearance, hotel check-in, examination and other scenarios, and can improve the efficiency and accuracy of verification.
Cooperation cases
Hangzhou Citizen Card
Real Estate Registration Center
Product scheme selection
Use offline SDK
Face recognition scenarios that cannot connect to the network, the network is unstable, the data security and privacy requirements are high, and the face database is small (usually less than 1w people). Such as access control machine, attendance machine, person identification machine, etc.
SDK+API offline and online mixed use
The network condition is good, the face database is small (below 1w) but changes frequently, and the business is often highly concurrent (such as attendance, cell gate). The offline face database handles some requests, and the API handles some requests, saving QPS resources.
Privatization deployment
The model service is deployed to the customer's local server and is applicable to the intranet environment with high security requirements, and the scene with a large face database (above 10000 levels, such as the millions of face databases for security). It is often used for large-scale and highly concurrent face recognition services.
Selection of living scheme
Passage scene
This scenario usually gives priority to the traffic speed to ensure that the traffic order and efficiency are not affected. Therefore, it is recommended not to use triple live detection, but only RGB or NIR live detection to ensure efficiency and safety.
Identity verification scenario
This scenario usually focuses on ensuring business security, and can provide more secure live solutions as far as possible. For example, RGB+NIR, or RGB+Depth, or even RGB+NIR+Depth, ensures a high rejection rate of attacks to the greatest extent.
Strong/dark scenes
In the scene with strong or weak light, RGB live bodies will be seriously affected. It is recommended to use NIR or Depth live bodies. At the same time, try to avoid the light in these two situations through product strategies, such as adding fill lights, equipping with light masks, etc.
Product pricing
Pricing of face offline recognition SDK
Serial number authorization one by one
Batch Authorization by Application
Serial number authorization
Online networking activation is carried out by inputting the serial number, which is convenient for authorization management and secondary distribution, and easier to control the application range. It is suitable for a few gates and access control equipment delivered with the project. Purchasing Consulting>
Permanent authorization
Free upgrade
Exclusive discount for mass purchase
Purchase Options
Authorization of face offline recognition SDK
Independent step price
RMB 119 for 100-999
term of validity
permanent
Purchase quantity
individual
¥ 11900
Buy Now
Related product recommendation
Offline face collection SDK
Offline call face detection, face tracking, face collection and other capabilities to quickly acquire face images and ensure the quality of the acquired face images. Cooperate with the API interface to efficiently build face recognition applications in various scenes
Face privatization deployment package
The face recognition model service that can be deployed to the "local server" provides basic functions such as face detection, face comparison, face search, and live detection. It supports a mega face database and can achieve millisecond response. Face detection and recognition capability supporting wearing masks
Dumu series soft and hard combination products
It includes four products: face application suite, face capture machine, video analysis box and AI lens module. Integration of software and hardware to achieve in-depth optimization of algorithms and hardware, equipped with preset applications, which can be directly deployed or secondary developed
Technical capability
Voice technology
Character recognition
Face and Human Body
Image technology
Language and knowledge
video technique
View all AI capabilities
Scenario application
Cooperation consulting
Pre sales consultation
Fill in your business needs, and the exclusive account manager will contact you as soon as possible to provide one-on-one consulting services
After sales intelligent assistant
Intelligent diagnosis to quickly solve the use problem
Contact Sales
For more information, please call 400-920-8999 to 1
Experience AI
Web end to AI Competency Experience Center
The mobile terminal opens Baidu APP "Scan"