Scanned Human 3D Models & Rich Data
Empowering computer vision with ground‑truth, high‑quality human 3D data for ML and simulation.
We offer world's #1 cited commercial human 3D models.
real scanned humans · full-body · textured meshes · dressed · everyday poses
Biggest & Fastest Growing Database
Over 35,000 scans and more than 5,500 refined 3D models available today. New scans on a weekly basis.
Human Diversity
Scanned individuals with origins in 72 different countries, spanning ages 7 to 83 and a wide array of body sizes.
Highest Quality
Cutting-edge photogrammetry technology with 330-camera setup. Lifelike and high-quality 3D models.
Ground Truth
3D models derived from scans of real people, providing accurate data offering a verifiable real-world benchmark.
Extra Components
Demographic and health related metadata, multi-view images, body measurements, segmentation, prop scans and more.
GDPR Compliant
Compliance in alignment with Articles 6(1) and 9(2) to utilize personal and health-related metadata, specifically for ML.
Trusted by industry leaders
to fuel the technologies of tomorrow.
Used in hundreds of papers
to drive scientific research and innovation.
Enhanced Monocular 3D Human Reconstruction Using Depth Information
ANIM: Accurate Neural Implicit Model for Human Reconstruction from a single RGB-D image
CVPR 2024 | Paper on arXiv | Project Page
Customizable 3D Human Avatar Creation from Sparse-View Videos
NECA: Neural Customizable Human Avatar
CVPR 2024 | Paper on arXiv
Advanced 3D Clothing Segmentation and Dataset Development
CloSe: A 3D Clothing Segmentation Dataset and Model
3DV 2024 | Paper on arXiv | Visit Project Page
Photorealistic 3D Human Reconstruction from Monocular Images
Photorealistic Monocular 3D Reconstruction of Humans Wearing Clothing
CVPR 2022 | Paper on arXiv | Visit Project Page
Realistic 3D Human Pose Estimation and Dataset Development
AGORA: Avatars in Geography Optimized for Regression Analysis
CVPR 2021 | Paper on arXiv | Visit Project Page
High-Resolution 3D Human Shape Estimation from Images
PIFuHD: Multi-Level Pixel-Aligned Implicit Function for High-Resolution 3D Human Digitization
CVPR 2020 | Paper on arXiv | Visit Project Page
This is just a small excerpt of studies that have used our data.
The human data solution for Computer Vision
Product Overview
We provide a diverse range of products and datasets, including static scans, rigged models, and dynamic scans of people in motion. Our offerings cater to various project needs such as Monocular and Single-view 3D Human Reconstruction, 3D Human Pose Estimation, Clothing Modeling and Fitting, 3D Avatar Creation or Human-Object Interaction in 3D Scenes. Explore our Products below or contact us for more information.
Posed Datasets
static scans in everyday poses and clothing
Available Posed Datasets:
Bodyshape Datasets
static A-Pose scans in bodyshape clothing
Available Posed Datasets:
accross all captured individuals
captured in our photogrammetry facility
available accross our range of products
available as multi-view images or unedited 3D
"Renderpeople offers an extensive collection of high-quality 3D meshes of people"
- Henrique Jesus and Hugo Proença, University of Beira Interior
Looking for more? We offer data others don't!
Extra Components
Beyond high-quality 3D models, we provide a rich array of corresponsing extra data and features.
Reach out to us anytime to explore further details or request samples tailored to your needs.
Detailed Metadata
For every scanned individual we collect personal data and health information, as well as metadata about clothing, pose and activity. Among others:
origin of parents, place of birth; age; skin color, hair color, eye color, body height, weight, body fat, BMI. Find a full list of available metadata attributes here.
Multi View Images
Our photogrammetry scanner shoots more than 330 images from different angles for every single scan we perform.
We offer all these images, as JPGs or RAWs, not only for the scans from which we have created finished 3D model products, but for every single one of the more than 34,000+ scans we have performed.
Body Measurements
We offer over 250+ body landmarks, including girths, distances, volumes, and heights, calculated using advanced geometric computation and machine learning.
To ensure maximum accuracy, we use body scans in tight-fitting clothing, minimizing distortion in the collected measurement data.
SMPL Registrations
We offer, on request, the registration of our 3D human scans to the popular SMPL body model, providing a standardized and anatomically accurate representation.
This service enhances the usability of scans for animation, simulation, and analysis, ensuring consistent and detailed human body data.
Point Cloud Data
We provide dense 3D point cloud data for every scan, available in .XYZ or .PLY formats.
This detailed spatial data is ideal for applications in 3D modeling, object recognition, scene understanding, and feature extraction, offering versatility and accuracy for specialized computer vision and pattern recognition projects.
UHQ RAW Meshes
Our scanning pipeline delivers ultra high-resolution geometries, producing 'Raw Meshes' with 50 million polygons and beyond, capturing intricate anatomical details.
These unedited geometries provide unmatched detail and accuracy, making them ideal for demanding applications in 3D human reconstruction and medical imaging.
3D Body Labels
We offer 3D body labeling services, providing detailed annotations directly on the 3D models.
These labels define critical features and landmarks, essential for supervised machine learning. This precise annotation enhances training datasets, improving the accuracy of algorithms for pose estimation, segmentation, and biometric analysis.
Additional Poses
We capture over 35 unique poses/scans per individual, providing a much wider range of data than our ready-to-use datasets include.
This extensive variety is ideal for research requiring diverse human poses and variations. If you need specific poses or additional data, we're here to accommodate your needs.
Segmentation Masks
Some of our datasets include segmentation masks, essential for human recognition and pattern analysis. These masks identify distinct body parts, providing the detailed data needed for training algorithms in synthetic 3D human avatar creation. This enhances the accuracy and realism of digital models, supporting advanced research and development.
Isolated Object Scans
Some of our 3D scans feature objects, like mugs, smartphones, or food items, held by the models.
We also provide separate scans of each prop, enabling you to train algorithms to identify and manage these objects in the 3D models. This allows for the option to exclude or modify objects according to your specific needs.
"Renderpeople [...] is the most popular dataset used in dressed people reconstruction methods"
- Shanghai Jiaotong University
Stay updated!
Newsletter
Subscribe to the HumanDataset newsletter to be always up to date for new products and dataset extensions.