Multi‑Modal Anti‑Counterfeiting
- Product Value
- Product Functions
- Product Advantages
- Application Scenarios
- Customer Cases
.png){kind=small}
-
.png)
Multi‑Dimensional Detection & Three‑Dimensional Protection
7 cutting‑edge anti‑counterfeiting algorithms
Built‑in domestically pioneering acoustic‑signal liveness detection technology
Champion of International Deepfake Competition FF++, Runner‑up of International Voiceprint Competition VoxSRC
-
Self‑Learning & Continuous Optimization
Powerful self‑learning capability to handle unseen new attacks
Continuously optimize anti‑counterfeiting effects based on massive real‑world business data
-
Efficient Deployment & Instant Response
Standard interface packaging greatly shortens deployment time compared with traditional expert models
Realize zero‑day vulnerability protection and rapid response to emerging attacks
-
Reliable Usage & High Interpretability
Interpret recognition results in natural language for user‑friendly understanding and better attribution analysis
Improve accuracy and transparency of decision‑making
-
Diverse Scenarios & Wide Application
Verified by real financial institutions and supports large‑scale liveness detection calls
High comprehensive attack interception rate, suitable for various application scenarios
-
Multiple Modes & Flexible Access
Multiple access forms to meet demands of different scenarios
More deployment methods to support cross‑industry expansion
Typical application scenarios to safeguard business security
.png "Multi‑Modal Anti‑Counterfeiting - Online Real‑Name Authentication")
Online Real‑Name Authentication
Online facial identity verification is widely used in the financial industry. As facial attack methods keep emerging, anti‑counterfeiting algorithms act as a firewall for businesses in such scenarios.
- Fast verification with liveness detection within 1 second
- Defend against 30+ attack types including photos, videos and masks
- Easy integration via SDK, H5, APP and other methods
- Support secondary fraud detection for multiple anti‑counterfeiting algorithms by backend services
- 95%+ accuracy for liveness detection
Camera Hijacking Detection
Attackers gain control of cameras through certain means to monitor, record or tamper with video streams without authorization. Attacks usually occur at the physical layer, system layer (RT, framework, ROM) and network layer. The most common camera hijacking attacks mostly happen at the ROM layer.
- Adopt ROM hijacking detection algorithms for 10 detection items including Android emulators, customized ROM, iOS jailbreak, cloud phones and suspicious permissions
- Apply repeated frame detection to cope with multiple attack attempts using the same material
- Use steganography and transmission encryption mechanisms to defend against network hijacking attacks
Group Fraud Identification
For group/intermediary fraud scenarios such as agency handling, document packaging and identity impersonation applications, identify group crimes by background similarity, logo recognition, IP/GPS positioning and other methods.
- Hundreds‑of‑millions‑level background image database accumulation
- Comprehensively judge group fraud risks combining GPS and time factors
- High‑precision feature extraction model for the financial industry
Well‑developed practical SOPs with deep industry and business insights
-
Banking
Identity Verification
Application Anti‑Fraud
Joint Operation
Account Login
-
Consumer Finance
Identity Verification
Group Fraud Identification
Agency Rights Protection Institution Identification
Telephone Business Processing
-
Auto Finance
Identity Forgery
Document Forgery
Telephone Verification Anti‑Fraud
-
Public Security
Anti‑Telecom Fraud
Identity Verification
Group Fraud Identification
China Merchants Bank
China’s first joint‑stock commercial bank fully held by corporate legal persons
View Details
Dezhu Multi‑Modal Biometric Anti‑Counterfeiting & Security Solution Secures Billions of Transfer Funds for Customers
It intercepts diverse attacks including AI face swapping, portrait activation, video synthesis, high‑fidelity masks, ROM hijacking, ID card forgery, voice synthesis and more. It builds cross‑modal liveness detection capabilities to improve identity authentication accuracy. Leveraging a deep‑fake anti‑counterfeiting large model, it enhances detection accuracy by integrating micro‑intelligent language reasoning models and high‑precision visual models. Combined with cross‑modal liveness detection and large‑model technologies, the average interception rate of deepfakes reaches 98% and presentation attack interception reaches 99%. The interception rate of common attacks such as AI face swapping, photo activation and video re‑shooting is increased to 98%. It ensures the safe transfer of billions of customer funds and effectively reduces risks brought by facial attacks, with zero breaches to date.
China CITIC Bank
Top‑Tier Commercial Bank | Fortune Global 500
View Details
Deep‑Fake Interception Rate Increased to Over 90%
The bank put forward high compatibility requirements for facial recognition on H5 terminals, demanding greatly improved customer experience and shorter waiting time, while realizing real‑time interaction via real‑time frame transmission technology.
The project finally shortened customer verification time to 3‑5 seconds and reduced waiting duration; it supports compatibility with 18 mobile models and mainstream browsers to ensure stable operation in diverse environments. Meanwhile, it provides over 20 deepfake detection algorithms, raising the deep‑fake interception rate to over 90%.
-409.png "Multi‑Modal Anti‑Counterfeiting - Bank of Suzhou")
Bank of Suzhou
City Commercial Bank | Founded in 2010
View Details
Facial Recognition Security Defense Engine Safeguards Business Security
Cooperating with Zhongguancun Kejin, Bank of Suzhou built a facial recognition security defense engine to guard against emerging technical attack risks such as facial adversarial samples and expression manipulation. It empowers facial authentication needs across multiple internal business scenarios, including online real‑name authentication, credit granting, device binding, mobile number modification, password modification and more. It ensures the security of related facial‑based businesses, regulatory compliance, avoids complaints and disputes, and prevents heavy financial losses.
A Commercial Bank
A Domestic Joint‑Stock Commercial Bank | Fortune Global 500
View Details
Strongest Anti‑Counterfeiting Shield for Credit Risk Control High‑Risk Attack Interception Rate Raised to 98%
The cross‑modal liveness detection + anti‑counterfeiting large‑model project jointly developed by Zhongguancun Kejin and the bank has become the strongest anti‑counterfeiting shield in its biometric verification system, covering credit granting scenarios for high‑net‑worth customers.
Through facial and voiceprint combined verification anti‑counterfeiting technology, the multi‑modal anti‑counterfeiting large model increases the interception rate of high‑risk attacks such as AI face swapping, photo activation and video re‑shooting to 98%, doubles the security of the entire business system, and ensures the safe landing of over 100 million funds.
More Customers
-247.png)
您的账号体验有效期已结束