3D Fragments dataset from Iberian vessels
A complete 3D fragments from vessels in different formats (point cloud, meshes and voxels), useful to be used in training tasks for deep learning.
Description
Accurate completion of fragmented archaeological ceramics is a critical but time-consuming manual task. To overcome the challenges of physical manipulation, reduce material deterioration, and improve reconstruction quality, deep learning is proposed as a support tool.
The text details the generation of a 3D fragment dataset using 1,052 Iberian vessels. Each mesh was broken into 2 to 10 fragments, resulting in a dataset of 1,040,428 point clouds and triangle meshes (plus 187,257 voxelizations) for training the models.
Alfonso López, Antonio J. Rueda, Rafael J. Segura, Carlos J. Ogayar, Pablo Navarro, José M. Fuertes (Instituto Universitario de Investigación de Arqueología Ibérica, Universidad de Jaén, Spain)