Conference
Open access
Sep 2026
Transferable Attacks on Open-Vocabulary Video Instance Segmentation via Dual-Objective Triggers
The Dual-Objective Triggers (DOT) is presented, the first transferable attack on OV-VIS that simultaneously exploits the vision–language coupling and temporal coherence, and Phase-Guided Ad-versarial Training is introduced, which injects perturbations primarily in the phase spectrum while blending amplitudes with clean...
Ming-Hao Shou, Ke-Sen Wang, Tong Zhang et al.
· Proceedings of the Thirty-Fi... · 0 citations