FACE_GUARD detects potential threats in an authentication environment. You can use this product if you are unable to fully integrate ID Verification.
The Face Guard product solution has been renamed to Device Guard, but the product code remains unchanged. For more information, see Face Guard is now Device Guard.
Integration overview
To integrate FACE_GUARD, your application must have mature facial recognition algorithms and a threat policy management system. Contact your business manager to determine if this solution is suitable for your business scenario.
FACE_GUARD detects device threats on different clients. Each client platform has different requirements for the operating environment and permissions. For specific requirements, see the Usage Notes section in the integration document for your client.
The FACE_GUARD workflow collects information from a client device. It then sends this information to a server for threat detection. You can use the server-side query API to retrieve detailed threat tags.
This solution requires an application server and a client. The following time series chart shows the complete FACE_GUARD workflow.
As shown in the preceding figure, the complete FACE_GUARD flow consists of two stages:
Stage one: Data collection and reporting. This stage occurs only on the client.
NoteThe SDK automatically connects to the FACE_GUARD server. Therefore, the client must be able to connect to Alibaba Cloud servers.
Stage two: Result retrieval and processing. In this stage, the application server retrieves the detection result, processes it, and returns the final result to the client.
If you use a facial recognition service with FACE_GUARD, note the following two time points for making calls from the client:
Time point 1: Before the user starts facial authentication and before the camera is activated. You can call the FACE_GUARD SDK initialization API when the user enters the guide page, or earlier. The earlier you call the API, the better.
Time point 2: After the user completes the facial authentication scan. Ensure a time gap of at least 3 seconds between this point and Time point 1.
When you call getDeviceToken, pass the bizId (the ID for the current facial authentication) to bind the current deviceToken to your unique business authentication ID. When the server queries for the threat result, it can then pass the bizId to verify the association.