Google Scanned Objects: A High-Quality Dataset of 3D Scanned Household Items

By Laura Downs, Anthony Francis, Nathan P. Koenig, Brandon Kinman, R. Hickman, Krista Reymann, T. B. McHugh, Vincent Vanhoucke. IEEE International Conference on Robotics and Automation, 2022

785 citations recorded by the source corpus.

Interactive 3D simulations have enabled break-throughs in robotics and computer vision, but simulating the broad diversity of environments needed for deep learning requires large corpora of photo-realistic 3D object models. To address this need, we present Google Scanned Objects, an open-source collection of over one thousand 3D-scanned household items released under a Creative Commons license; these models are preprocessed for use in Ignition Gazebo and the Bullet simulation platforms, but are easily adaptable to other simulators. We describe our object scanning and curation pipeline, then provide statistics about the contents of the dataset and its usage. We hope that the diversity, quality, and flexibility of Google Scanned Objects will lead to advances in interactive simulation, synthetic perception, and robotic learning.

Subfields: Computational Mechanics, Computer Graphics and Computer-Aided Design, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/9f5120b815fddaaef25c7042035ffe5680507a65

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computational Mechanics

In Computer Graphics and Computer-Aided Design

In Computer Vision and Pattern Recognition

Citations in the corpus

Wonder3D: Single Image to 3D Using Cross-Domain Diffusion (2023, methodology): “The qualitative comparisons with baseline methods on GSO [14] dataset in terms of the reconstructed textured meshes.”
Splatter Image: Ultra-Fast Single-View 3D Reconstruction (2023, methodology): “Compared to these slower baselines, Splat-ter Image is not only competitive in terms of quality, but in fact state-of-the-art in several cases, improving both reconstruction PSNR and LPIPS.”
VIMA: General Robot Manipulation with Multimodal Prompts (2022, background): “Bowl and pan are from Google Scanned Objects (Downs et al., 2022) while others are from Ravens (Zeng et al.”
InstantMesh: Efficient 3D Mesh Generation from a Single Image with Sparse-view Large Reconstruction Models (2024, methodology): “We evaluate the quantitative performance us-ing two public datasets, i.e ., Google Scanned Objects (GSO) [10] and OmniObject3D (Omni3D) [54].”
GS-LRM: Large Reconstruction Model for 3D Gaussian Splatting (2024, methodology): “We evaluate our model on two 3D object datasets including the full Google Scanned Objects (GSO) [21] that contains 1009 objects and the Amazon Berkeley Objects (ABO) [17] dataset from which we sample 1000 objects.”
ProcTHOR: Large-Scale Embodied AI Using Procedural Generation (2022, background): “Additionally, we plan to scale up our asset databases by leveraging many open-source 3D asset databases, such as ABO [26], PartNet [81], ShapeNet [15], Google Scanned Objects [30], and CO3D [97], among others.”
FoundationPose: Unified 6D Pose Estimation and Tracking of Novel Objects (2023, methodology): “We developed a novel synthetic data generation pipeline for training, powered by the recent emerging resources and techniques: large scale 3D model database [6, 10], large language models (LLM), and diffusion models [4, 23, 49].”
GRM: Large Gaussian Reconstruction Model for Efficient 3D Reconstruction and Generation (2024, methodology): “We use Google Scanned Objects (GSO) [20], and render a total of 64 test views with equidistant azimuth at { 10 , 20 , 30 , 40 } degree elevations.”
Objaverse: A Universe of Annotated 3D Objects (2022, background): “Slightly larger datasets of photo-realistic objects include GSO [19], PhotoShape [53], ABO [12] and 3D-Future [22], and ShapeNet [7] with object counts in the tens of thousands, see Fig.”
PointOdyssey: A Large-Scale Synthetic Dataset for Long-Term Point Tracking (2023, methodology): “Similar to Kubric [24], we populate out-door scenes with random rigid objects from GSO [19] and PartNet [42].”
LRM: Large Reconstruction Model for Single Image to 3D (2023, neutral): “To evaluate the performance of LRM on arbitrary images, we collected novel images from Ob-javerse (Deitke et al., 2023), MvImgNet (Yu et al., 2023), ImageNet (Deng et al., 2009), Google Scanned Objects (Downs et al., 2022), Amazon Berkeley Objects (Collins et al., 2022), captured new images in the r”
SyncDreamer: Generating Multiview-consistent Images from a Single-view Image (2023, methodology): “We report PSNR, SSIM [76], LPIPS [90] and reconstructed point number by COLMAP [57] on the GSO [16] dataset.”
Zero-1-to-3: Zero-shot One Image to 3D Object (2023, methodology): “We evaluate both tasks on Google Scanned Objects (GSO) [11], which is a dataset of high-quality scanned household items, as well as RTMV [51], which consists of complex scenes, each composed of 20 random objects.”
Qwen-Image Technical Report (2025, neutral): “Second, to evaluate the model’s spatial understanding and generative capability in 3D vision tasks, we benchmark its performance on novel view synthesis (Downs et al., 2022) and depth estimation (Bochkovskii et al., 2024).”
SV3D: Novel Multi-view Synthesis and 3D Generation from a Single Image using Latent Video Diffusion (2024, methodology): “Frame 1 Frame 20 We evaluate the SV3D synthesized multi-view images on static and dynamic orbits on the unseen GSO [10] and OmniObject3D [54] datasets.”
One-2-3-45: Any Single Image to 3D Mesh in 45 Seconds without Per-Shape Optimization (2023, methodology): “We also quantitatively compare the approaches on Objaverse [11] and GoogleScannedObjects (GSO) [14] datasets.”
VGGT: Visual Geometry Grounded Transformer (2025, neutral): “7, despite not requiring the input camera parameters and using less training data than LVSM, our model achieves competitive results on the GSO dataset [17].”
Instant3D: Fast Text-to-3D with Sparse-View Generation and Large Reconstruction Model (2023, methodology): “We do the comparisons on the Google Scan Object (GSO) dataset (Downs et al., 2022), which consists of 1019 objects.”