HSSD: 211 Human-Authored 3D Scenes with 18,000+ Object Models

211 human-authored 3D scenes and over 18,000 models of real-world objects comprise the Habitat Synthetic Scenes Dataset (HSSD). This dataset is designed to more closely mirror real interiors than prior synthetic scene collections. The dataset was created by 'hssd' and was last updated on the Hugging Face platform in June 2023.

Use Cases

• Training embodied AI agents for navigation and interaction based on the dataset's diverse 3D scenes.
• Benchmarking scene understanding and object recognition models using the 18,000+ real-world object models.
• Developing and testing robotics simulation environments based on the human-authored, realistic interiors.
• Generating synthetic training data for computer vision tasks based on the detailed 3D scene representations.

Strengths

• Contains 211 distinct 3D scenes, providing a substantial collection for simulation.
• Includes over 18,000 models of real-world objects, offering significant diversity for training.
• Scenes are human-authored to more closely mirror real interiors than prior datasets.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and file formats are unknown, which may limit suitability assessment.
• Last updated 2023-06-15; freshness should be verified for current projects.

Provenance

Source

hssd

Collection Method

Human-authored creation of 3D scenes and object models.

Time Range

null

Freshness

Last updated 2023-06-15 03:32:23

Geography

null