Real Human Match

Contents

Visualization of the shapes in the dataset with color-coded correspondences:
colored regions indicate accurate correspondences, grey areas indicate no accurate correspondences available.

• original_shapes/: Original textured 3D meshes sourced from the 4DHumanOutfit dataset.
• shapes/: 3D mesh files in OFF format.
• shapes_template/: SMPL (Skinned Multi-Person Linear) model fittings corresponding to each mesh in shapes/.
• shapes_aligned/: The same shapes as in ‘shapes/’, but rigidly aligned.
  • All meshes are reoriented to a consistent coordinate system:
    • Front-back and right-left alignment.
    • Y-axis points up; Z-axis points forward.
  • All meshes are upright and facing the same direction for consistent evaluation.
• ground_truth/: Ground truth correspondences.
  • For each vertex in a mesh from the ‘shapes’ directory, this specifies the corresponding vertex index on the SMPL template.
  • A value of -1 indicates that ground truth correspondence is not available for that vertex.
• geodist_template/: Precomputed geodesic distance matrices on the SMPL template shapes, used for evaluating matching accuracy and quality.
• evaluation_script.py: Script for evaluating the correspondences computed on the dataset’s shapes, compares predicted correspondences with ground truth using geodesic error metrics.

Downloads

The data is released for non-commercial research purposes. The release is subject to a license that includes compliance with the General Data Protection Regulation (GDPR) of the European Union, and requests need to be signed by the legal representatives of the interested institutions.

To obtain this authentification, please contact us and provide information on the name and address of your research institution, the name, title and e-mail address of the legal representative of your institution. If your institution is within the European Union, we further require the e-mail address and postal address of the Data Protection Officer (DPO) of your institution.