Testing the competition¶
In 2021 I tested some liveness implementations from the competition and showed how vulnerable they were. You can find the report at https://www.doubango.org/SDKs/face-liveness/docs-deprecated/Testing_the_competition.html. I’ll not do the same today (2025) as the exercise was time consuming. In the previous sections we presented how our application can detect image and video injections. To ensure that the methods used to test our injection detection implementation are effective I had to try them against other implementations. So, I found 1 application for each injection type (1 for video and 1 for image).
Video injection (Regula Forensics)¶
The method used to inject the deepfake is described at this section. I have tested both their active and passive liveness implementations.
Implementation: https://faceapi.regulaforensics.com
Test done on: 2025-08-16 03-20-49 CEST.
Obama’s face used as deepfake: Obama.jpg.
Image injection (kby-ai)¶
Image injection is done using Android 13 on BlueStaks emulator and OBS Studio Virtual Camera.
Implementation: https://play.google.com/store/apps/details?id=com.miniai.liveness
Test done on: 2025-08-16 07-01-39 CEST.
Macron’s face used for injection: Macron.jpg
