SHREC'20: Shape correspondence with non-isometric deformations - AIM@SHAPE (data)

We have curated a dataset of 3D scans of objects from existing data repositories. All scans are of quadruped mammals. Shapes were primarily created using a variety of capturing techniques. Ground-truth corresdpondences were then established by three experts. Shapes (e.g., a giraffe and a dog) exhibit substantial non-isometric deformation, which prove challenging for existing shape matching methods. The dataset consists of both partial and watertight (i.e., full) meshes. Additional complexities caused by scanning, such as geometric and topological change, are also present. Pairs of shapes are assigned to test-sets of gradually increasing non-isometry.

This dataset has been split into two parts to help make the licences clearer. This part contains the following shapes from the AIM@SHAPE-VISIONAIR Shape Repository: elephant_b, giraffe_b, cow, camel_a, camel_b, bear.

^*When downloading these models you must still agree to the AIM@SHAPE General License for Shapes.

⁺These models are also available from another source (http://www.wisdom.weizmann.ac.il/~vision/indexing/data.html)

The Licence terms for these data are available at http://visionair.ge.imati.cnr.it/ontologies/shapes/download.jsp?id=803-Neptune_-_clean__watertight__4M_triangles_&classid=ManifoldSurfaceMesh&groupid=&type=original