Abstract: We target a 3D generative model for general natural scenes that are typically unique and intricate. Lacking the necessary volumes of training data, along with the difficulties of having ad ...
Abstract: We propose a general attack framework based on evolutionary algorithms to quickly and efficiently generate low-perturbation adversarial samples for 3D point cloud data. Specifically, we ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results