Anmelden
Einblick - Video adversarial attack
Attacking Video Recognition Systems via Adversarial Logo Style Transfer
A novel attack framework, LogoStyleFool, is proposed to fool video recognition systems by superimposing a stylized logo on the input video, achieving superior attack performance and semantic preservation compared to existing patch-based attacks.
Efficient Black-box Video Adversarial Attacks via Style Transfer
StyleFool, a black-box video adversarial attack framework, leverages style transfer to efficiently fool video classification systems with unrestricted perturbations.
Produkte | Ressourcen
© 2024 by Linnk AI