TransPhy3D: RGB Video Sequences with Merged 3D Scene Reconstructions

TransPhy3D is a grouped dataset built from a random subset of the TransPhy3D test split. Each group contains one RGB video sequence paired with a merged 3D reconstruction of the same scene. The dataset was created by harpreetsahota and last updated on May 22, 2026.

Use Cases

• Train models for 3D scene reconstruction based on paired RGB video sequences.
• Evaluate the performance of video-to-3D reconstruction algorithms.
• Benchmark multi-view geometry algorithms using synchronized video and 3D data.
• Develop multimodal models that fuse temporal video information with spatial 3D structure.

Strengths

• Dataset groups provide synchronized RGB video and 3D reconstruction data for the same scene.
• Data is derived from the TransPhy3D test split, suggesting a structured evaluation benchmark.

Limitations

• Dataset size, row count, and file formats are unknown.
• Column-level documentation is absent; field semantics must be inferred after download.
• The dataset is a random subset, which may not represent the full distribution of the original test split.

Provenance

Source

huggingface

Collection Method

Built from a random subset of the TransPhy3D test split.

Freshness

Last updated 2026-05-22 18:06:25.