1 paper · 1 filter
Jian Wang, Baoyuan Wu, Li Liu +1
The rapid evolution of generative AI has increased the threat of realistic audio-visual deepfakes, demanding robust detection methods. Existing solutions primarily address unimodal…