Abstract: Existing adversarial attacks for face anti-spoofing predominantly assume that the model parameters are fully known, and often overlook the transferability of adversarial examples across ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results