The prosperity of deep neural networks (DNNs) is largely benefited from
open-source datasets, based on which users can evaluate and improve their
methods. In this paper, we revisit backdoor-based dataset ownership
verification (DOV), which is currently the only feasible approach to pro