A lot of fun at CVPR this year. Hosted the Authenticity & Provenance in gen AI (APAI) workshop, alongside presenting 3 new papers from our lab (linked below): 1) Hany Farid's and my work on detecting real vs. fake explosions using spatiotemporal dynamics; 2) perceptual study showing video length doesn't improve human ability to spot fakes, led by Catherine Davodi; and 3) the DeepSpeak dataset alongside Maty Bohacek. It was wonderful seeing so many new and familiar faces!
Explosions - https://lnkd.in/gDB5fR4f
Perceptual study - https://lnkd.in/g6z2Q3ya
DeepSpeak - https://lnkd.in/gkfR6hrV
