heig-vd-geo/scop-datasets

Photogrammetry and LiDAR datasets (HEIG-VD)

Datasets from the Geomatics group of HEIG-VD (Yverdon-les-Bains, Switzerland) for photogrammetry and LiDAR: image orientation, bundle adjustment, camera calibration, orthophotos and dense point clouds. They are used as benchmarks for the SCOP software but do not depend on it: each dataset is a folder of ordinary files (images, coordinates, projects).

The repository also holds a catalogue of external public datasets (catalog.yaml, folder external/): link to the original source, checked licence, citation and download script. Those datasets are not copied here.

Our datasets (datasets/)

Licence: CC BY 4.0. Please cite "HEIG-VD, Geomatics group" and this repository.

Dataset	Sensor	Date, place	Contents
drone_heig_2019	senseFly S.O.D.A. (eBee)	16.07.2019, Yverdon (HEIG-VD)	32 geotagged drone images (small demo set), 31 ground points (LV95) with image measurements
sequoia_champs	Parrot Sequoia (4 bands + RGB)	21.07.2019, La Sarraz – Cossonay area (VD)	multispectral, 116 captures (4 bands + RGB)
p4rtk_mezieres	DJI Phantom 4 RTK	08.06.2020, Mézières (VD)	263 RTK images, Metashape project
gopro360	GoPro 360 dual fisheye	HEIG-VD, classroom	139 fisheye pairs, measured rig, 8 ground control + 2 check points (LV95)
calib_atelier	Nikon Z5 II, 28 mm	26.05.2026, Yverdon	calibration polygon, 16 points in LV95/RAN95
calib_j06c_16mm	Nikon Z5 II, Viltrox 16 mm	10.06.2026, HEIG-VD room J06C	1600 targets in a local frame, Metashape projects
mont_blanc	Sony α7R IV, 24 mm	25.09.2024, HEIG-VD	plastic relief map, 5 images, scale bars, Metashape project (dense cloud, mesh, DEM)
banc_stereo_voute	2 × The Imaging Source monochrome cameras	12.2021, laboratory	low-cost stereo rig, 59 synchronised pairs of a masonry vault, 8 targets (local frame), Metashape 4D project
j07_leica	Leica terrestrial laser scanner	07.10.2024, HEIG-VD (J07)	5 indoor TLS stations (E57, ~10 M points each, colour), Leica registration poses, panoramas
riegl_paire	Riegl VZ-2000i	21.04.2021, Yverdon	pair of TLS stations (PLY, 4.5 M points each) with the reference rigid transformation, for registration
p4rtk_yverdon_2020	DJI Phantom 4 RTK	07.01.2020, Yverdon	111 RTK images, drone RINEX + MRK, YVD2 base RINEX for PPK, 25 ground points (LV95) with image measurements

Each folder has its own README.md: sensor, reference frame, ground control, expected accuracy.

External public datasets (external/)

One folder per dataset, with a download.sh (curl, size and md5 check, extraction, resume of an interrupted download) or, when access requires a registration, a README.md with the procedure.

./external/eth3d_facade/download.sh --dest ~/data/eth3d_facade        # default files
./external/vineyard_botrytis/download.sh --all --dest ~/data/vineyard  # + optional files

The script prints the licence and the citation. For a dataset that does not allow commercial use, or has no declared licence, it asks you to accept its terms (ACCEPT_LICENCE=1 to skip the question).

Dataset	Type	Licence	Commercial use
odm_helenenschacht	drone	GPL-3.0	yes
odm_waterbury	drone	not declared	—
odm_ziegeleipark	drone (29.8 GB)	CC0 1.0	yes
vineyard_botrytis	multispectral drone	CC BY 4.0	yes
pix4dmatic_example_100	drone	Pix4D terms (training)	no
eth3d_courtyard, _delivery_area, _facade, _meadow, _pipes	terrestrial + reference scan	CC BY-NC-SA 4.0	no
colmap_gerrard_hall, _south_building, _person_hall	terrestrial	not declared	—
middlebury_temple_ring	object	not declared	—
aachen_day_night_v1_1	city, localisation	CC BY-NC-SA 4.0	no
dbat_stpierre	terrestrial (PhotoScan project)	BSD-like, citation requested	yes
geco_heritage_images_pointclouds	heritage, images + semantic point clouds	CC BY 4.0	yes
ch_photoslam3d	heritage, drone + terrestrial + SLAM	CC BY 4.0	yes
isprs_dortmund	drone, terrestrial, aerial, TLS	ISPRS/EuroSDR registration	no
dn_lidar	ALS + trajectories	not declared	—
cherish, cyark_open_heritage_3d	TLS + UAV, heritage (spotted)	CC BY 4.0 / CC BY-NC 4.0	yes / no

"Not declared": the original repository gives no licence, so no right of redistribution or commercial use can be assumed. Licences and links checked in October 2026 (field checked of the catalogue).

Catalogue

catalog.yaml describes each dataset: identifier, type, sensor, licence (SPDX), citation, DOI, files (URL, size, md5, extraction, optional or not) and where the images, ground control and reference are after extraction. The external/*/download.sh scripts are generated from it by tools/make_download_scripts.py.

Citation

HEIG-VD, Geomatics group (2026). Photogrammetry and LiDAR datasets. Hugging Face, https://huggingface.co/datasets/heig-vd-geo/scop-datasets

For an external dataset, cite its source (field citation of the catalogue).