Skip to content

C3VD colonoscopy scenes as Gaussian splats and point clouds (EndoGSLAM fits from PR-ENDO)

pl.load("c3vd_endoscopy_splats") loads this dataset. Ten colonoscopy sequences of the C3VD phantom dataset reconstructed with EndoGSLAM and released with PR-ENDO (relightable Gaussian splatting for endoscopy): per scene the optimised Gaussian splat (point_cloud.ply with the 3DGS attributes), the Gaussian centres (means3D.ply) and the per-step parameter dumps (.npz / .pt). Read out of the 7.8 GB zip on Zenodo by HTTP Range. The frame is the scene's own camera frame; the scale follows the C3VD phantom (millimetres, TO CONFIRM).

import pardesline as pl

ds = pl.load("c3vd_endoscopy_splats")
print(ds.describe())  # roles, frames, units, labels, licence, citation
s = ds[0]
obj = s.splat  # fetched now, cached under ~/.cache/pardesline/datasets/c3vd_endoscopy_splats

Provenance and licence
Source remote archive read by HTTP Range: https://zenodo.org/records/15732143/files/C3VD_endogslam_optimized.zip
Scheme archive
Size of the whole source 7.7 GB
Licence CC-BY-4.0
Redistribution unknown
Licence note CC BY 4.0 on the PR-ENDO Zenodo record (checked 2026-10-14); the underlying C3VD dataset is CC BY-NC-SA 4.0, so commercial use is TO CONFIRM.
Terms https://zenodo.org/records/15732143
Homepage https://github.com/SanoScience/PR-ENDO
Cite Kaleta J., Dall'Alba D., Plotka S., Korzeniowski P. PR-ENDO: Physically Based Relightable Gaussian Splatting for Endoscopy. arXiv:2411.12510, 2024. Bobrow T.L., Golhar M., Vijayan R., et al. Colonoscopy 3D Video Dataset with Paired Depth from 2D-3D Registration. Med Image Anal 90:102956, 2023.
Samples grouped by scene

Roles (what each sample contains)

Role Type Frame Units Labels / layout Pattern or file
splat splat sensor mm C3VD_endogslam_optimized/{scene}/point_cloud/iteration_*/point_cloud.ply
Optimised 3-D Gaussian splat (PLY with position, scale, rotation, opacity, SH colour)
points points sensor mm C3VD_endogslam_optimized/{scene}/means3D.ply
Gaussian centres as a plain point cloud

Standardized layout

Files are fetched on first access into the dataset cache and stay in their original format; ds.export(folder) writes every sample as standard files (PLY for points and meshes, MetaImage for volumes and label maps) with a .pardesline.json sidecar carrying frame, units, labels, licence and provenance, plus a manifest for the folder. Label maps become exact surfaces with to_mesh(label=k), DICOM series become one volume, raw arrays are decoded through the layout above.

ds = pl.load("c3vd_endoscopy_splats").take(2)
ds.export("./standardized")        # PLY / MetaImage + sidecars + manifest.json

Algorithms and models that run on it

Generated from src/pardesline/data/manifests/c3vd_endoscopy_splats.json on 2026-10-14.