BaliMask3D dataset for 3D completion and reconstruction of traditional Balinese masks.

Abstract

The BaliMask3D dataset provides high-quality 3D models of traditional Balinese masks to support 3D completion and reconstruction tasks, addressing the need for cultural heritage preservation. Digitized using 360-degree photogrammetry, the dataset comprises 27 classes of masks selected from museums in Bali, representing diverse and intricate carvings. The dataset includes key processing stages: preprocessing, training with VQ-VAE, SDFusion, and evaluation, enabling the restoration of incomplete or damaged 3D artifacts. This dataset serves as a valuable resource for developing and benchmarking deep learning techniques in 3D reconstruction while fostering advancements in cultural preservation and computer vision research.