OmniRooms: Synthetic Indoor Panoramas for 3D Perception

OmniRooms is a large-scale synthetic indoor dataset for panoramic 3D perception, depth estimation, and monocular view synthesis. It contains 16 large indoor scenes with multiple rooms, totaling 271,000 equirectangular RGB panoramas. The dataset was created by Insta360-Research and was last updated in June 2026.

Use Cases

• Training models for panoramic 3D perception based on equirectangular RGB panoramas.
• Developing depth estimation algorithms from paired panoramic images.
• Researching monocular view synthesis in complex indoor environments.
• Exploring 3D Gaussian Splatting techniques from universal camera inputs.

Strengths

• Contains 271,000 equirectangular RGB panoramas, indicating a substantial volume of data.
• Covers 16 large indoor scenes with multiple rooms, suggesting diversity in environments.
• Specifically designed for research on 3D reconstruction and 3D Gaussian Splatting.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data is synthetic, which may not fully capture the complexity of real-world scenes.

Provenance

Source

Insta360-Research

Collection Method

Synthetically generated.

Freshness

Last updated 2026-06-05 17:33:55; freshness should be verified.