The Cost-Effective Face Liveness Detection in Real-World Applications
Unlocking security and innovation: the cost-effective journey of liveness detection in real-world applications
Situation: Imagine a small fast-food network restaurant where you can place an order yourself at a special terminal. And before payment, the customer is offered to get his personal discount on the loyalty program just by smiling at the camera. What could go wrong?
Problem: Dishonest customers quickly learned to deceive the system, showing photos of people from the internet (including social networks) and taking a large discount for themselves when ordering.
Possible solution: Enable liveness check during face authorization. This would eliminate the simplest presentation attack - the smartphone screen attack (By the way, Biometric Anti-fraud handles this well).
As a result, face recognition to receive a discount had to be rolled back and returned to traditional SMS codes. According to rough estimates, the network managed to lose about $120,000 during the operation of the mechanism.