Aria Synthetic Indoor Scenes for 3D Understanding

100,000 procedurally-generated indoor scenes comprise this synthetic dataset. It was created by projectaria for research on 3D scene understanding, object detection, and tracking, with a last update in September 2024. The dataset simulates sensor data matching the characteristics of Project Aria glasses.

Use Cases

• Training 3D object detection models using procedurally-generated scene geometry and object placements.
• Developing and benchmarking visual tracking algorithms on simulated sensor streams from Project Aria glasses.
• Researching scene understanding tasks like layout estimation from the synthetic indoor environments and their populated objects.
• Simulating and testing sensor fusion approaches with the dataset's multi-modal synthetic data matching real device characteristics.

Strengths

• Contains 100,000 unique procedurally-generated indoor scenes.
• Features realistic 3D objects and simulated sensor data mirroring a specific hardware platform.

Limitations

• The dataset is entirely synthetic, which may introduce a sim-to-real gap for model deployment.
• Specific details on scene diversity, object categories, and data volume per scene are unknown.

Provenance

Source

projectaria

Collection Method

Procedurally generated synthetic data.

Freshness

Last updated on 2024-09-21.