R3PM-Net: Synthetic and Real-World Industrial Point Cloud Datasets

Two datasets, Sioux-Cranfield and Sioux-Scans, proposed in the R3PM-Net paper to address the gap between synthetic and real-world industrial data. The repository includes pickle files from a subset of the Sioux-Cranfield dataset for model training. The dataset was authored by YasiiKB and last updated on Hugging Face in April 2026.

Use Cases

• Training 3D object recognition models based on synthetic and real-world industrial point clouds.
• Benchmarking sim-to-real transfer learning methods based on the paired dataset structure.
• Developing point cloud registration or reconstruction algorithms based on the Sioux-Scans data.
• Evaluating model robustness using the provided down-sampled ModelNet40 subset.

Strengths

• Explicitly designed to address a known research gap between synthetic and real-world industrial data.
• Includes pre-processed pickle files for a subset of the Sioux-Cranfield data, facilitating immediate use for training.
• Repository structure is documented, listing specific dataset archives (sioux_cranfield.zip, sioux_scans.zip).

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count, file formats, and dataset size are unknown, which may limit suitability assessment.
• License information is unavailable, restricting clarity on permissible use.

Provenance

Source

Hugging Face dataset repository authored by YasiiKB, associated with the R3PM-Net paper.

Collection Method

Likely contains both synthetically generated and real-world scanned industrial point cloud data.

Freshness

Last updated 2026-04-09 07:15:57; freshness should be verified.