Fig. 9From: Deep 3D mesh watermarking with self-adaptive robustnessVisual quality on remeshed Asiadragon (top two rows) and remeshed Hand (bottom two rows) dataset. a Original; b Model; c Model\(\dag\); d Pre-Model; e Pre-Model\(\dag\). \(\dag\) denotes training without the attack layers, Model represents the model trained on the remeshed dataset, and Pre-Model represents the model trained on the original datasetBack to article page