VolHuMe: a High-Resolution Large Scale Dataset of Volumetric Human Meshes
2026-06-22 • Graphics
GraphicsComputer Vision and Pattern Recognition
AI summaryⓘ
The authors created VolHuMe, a new dataset with very detailed 3D scans of 104 people using many cameras to capture their bodies and faces from multiple angles. This dataset includes high-quality 3D models, images, and detailed parts like hands and faces. Their setup focuses on capturing close-up details better than previous collections, which helps improve how accurately people can be reconstructed in 3D and 4D. They tested current methods on their dataset to show its usefulness and to highlight where existing tools still struggle.
4D human scansvolumetric studioSMPL-Xhigh-resolution meshesmulti-view images3D reconstructiondepth camerasgarment segmentationpoint cloudsrigged meshes
Authors
Giulia Martinelli, Niccolò Bisagno, Nicola Garau, Esa Rahtu, Nicola Conci
Abstract
We introduce VolHuMe, a dataset of high-quality 4D human scans captured with a state-of-the-art volumetric studio using 64 RGB and 32 depth cameras. VolHuMe contains individual captures of 104 subjects and provides extensive ground truth, including SMPL-X, high-resolution meshes, multi-view RGB/depth images, rigged meshes, point clouds, garment segmentation, and detailed hand and facial geometry. Unlike prior datasets that primarily rely on full-body imagery, VolHuMe uses a close-range, high-resolution capture setup that preserves fine-grained body-part details, improving geometric fidelity and texture resolution. We benchmark VolHuMe on state-of-the-art methods across 3D and 4D human reconstruction tasks, showcasing the dataset's quality and exposing the limitations of current evaluation testbeds.