Sign in to view source links and access this dataset
Description
A collection of open-source 3D assets for simulation, reverse-modeled at 1:1 scale to real-world objects. The dataset includes articulated objects with joints, physics parameters, semantic labels, and PBR materials, created by Extwin and last updated on 2026-06-19.
Use Cases
Training reinforcement learning agents for robotic manipulation based on articulated objects with joints.
Developing computer vision models for object detection and segmentation based on assets with semantic labels.
Building and testing physics-based simulations based on objects with defined physics parameters.
Creating synthetic training data for autonomous systems based on high-fidelity 3D models.
Strengths
Assets are reverse-modeled at 1:1 scale to real-world objects to ensure quality.
Most objects include joints and follow a unified naming convention.
Assets include physics parameters, semantic labels, and PBR materials.
Limitations
Description metadata is limited; actual data quality requires manual inspection after download.
Column-level documentation is absent; field semantics must be inferred after download.
Row count is unknown, which may limit suitability assessment.
Provenance
Source
Extwin via huggingface
Collection Method
Reverse-modeled from real-world objects.
Freshness
Last updated 2026-06-19 09:28:34; freshness should be verified.
License is unknown; users must verify terms before use.