Keep DRÆMing: Discriminative 3D anomaly detection through anomaly simulation

Vitjan Zavrtanik, Matej Kristan, Danijel Skocaj

Abstract: Highlights•A novel 3D anomaly simulation process is proposed.•Simulation enables the training of discriminative methods for 3D anomaly detection.•The 3DRÆM anomaly detection method demonstrates the simulation process efficiency.•3DRÆM achieves state-of-the-art results on the MVTec3D benchmark.