ARKit Labelmaker: Dense Semantic Annotations for Indoor 3D Scenes

ARKit Labelmaker is a dataset that provides dense semantic annotations for the ARKitScenes dataset, automatically generated at scale. It is described as the first large-scale, real-world 3D dataset with dense semantic annotations. The dataset was created by 'labelmaker' and was last updated on the Hugging Face platform in March 2025.

Use Cases

• Training 3D semantic segmentation models based on the dense semantic annotations.
• Benchmarking model performance on ScanNet and ScanNet200 datasets based on the auto-generated training data.
• Advancing research in large-scale, real-world indoor 3D scene understanding based on the dataset's scale and annotation density.

Strengths

• Provides the first large-scale, real-world 3D dataset with dense semantic annotations.
• Annotations are automatically generated, enabling creation at scale.
• Training on this data is reported to push state-of-the-art performance on established benchmarks like ScanNet and ScanNet200.

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, file formats, and license information are unknown, which may limit suitability assessment.

Provenance

Source

labelmaker (author), likely building upon the ARKitScenes dataset.

Collection Method

Dense semantic annotations are automatically generated at scale.

Freshness

Last updated 2025-03-25 13:46:55; freshness should be verified.