HovoPhoto40 – A high-resolution multi-epoch photogrammetry dataset for geomonitoring of the Mt. Hochvogel (5 epochs, 40 images each)

The HovoPhoto40 dataset provides a high-resolution multi-epoch photogrammetric image dataset for alpine geomonitoring at the summit area of the Hochvogel mountain (Northern Limestone Alps, Austria/Germany). The dataset documents progressive rock slope deformation across a major summit crevice separating a stable northwestern block from an unstable southeastern block. The dataset consists of five photogrammetric acquisition epochs conducted in 2018, 2021, 2022, 2023, and 2024. For each epoch, 40 manually selected images were acquired along a handheld image strip using a Sony Alpha 7R II camera (42 MP, 25 mm focal length). Image regions containing persons have been blurred to protect privacy. The average camera spacing is approximately 0.75 m, resulting in a ground sampling distance of approximately 2 mm at an object distance of about 10 m. For each epoch, the dataset provides: Original image data (YYYY_images.zip) Undistorted image versions (YYYY_images_undist.zip) Internal and external camera orientation parameters (in YYYY_param.zip) Projection matrices and camera model (in YYYY_param.zip) Manually marked image (2D) and measured object (3D) coordinates of ground control and reference points (in YYYY_param.zip) Photogrammetric dense point clouds (YYYY_HV40_point_cloud.ply/.las) The parameter files include both measured 2D image coordinates and corresponding 3D object coordinates of control points (HV40_YYYY_image_points.txt, HV40_YYYY_object_points.txt). A subset of reference points labeled R1, R2, R3, R5, R6, and R7 is available in all epochs. These points were repeatedly measured using total station observations and represent the displacement of the unstable mountain area, serving as reference values for change analysis. All other labeled points are typically only present within individual epochs and were assigned arbitrary identifiers. Ground control and reference points were measured using a hybrid geodetic monitoring network combining tachymetric observations and GNSS baselines. For all points, an accuracy of approximately 4 mm per coordinate is assumed. The number of ground control points varies between epochs due to accessibility and field conditions. All photogrammetric processing, including bundle adjustment, camera calibration, image undistortion, and dense 3D reconstruction, was performed using Pix4Dmapper. Detailed information about processing conventions, coordinate system definitions, and parameter formats can be found in the Pix4D documentation. Object points, dense point clouds, and external camera orientation parameters are provided in a local Cartesian XYZ coordinate system defined during photogrammetric processing. For transformation into a global reference system, a constant offset of Δ=[32608000, 5248000, 2600] can be added to all object coordinates if required. For internal computations (e.g., reprojection of 3D object points into image space using the provided projection matrices), an epoch-specific coordinate offset (HV40_YYYY_offset.xyz) exported by Pix4D must be applied. Example Python scripts are included in the dataset demonstrating (i) the conversion of camera poses into COLMAP format and (ii) the reprojection of 3D object points into the corresponding images. The internal orientation includes focal length, principal point coordinates, as well as radial and tangential distortion coefficients (HV40_YYYY_pix4d_calibrated_internal_camera_parameters.cam). The external orientation is given by the projection center coordinates and the corresponding orientation angles as omega, phi and kappa (HV40_YYYY_calibrated_external_camera_parameters.txt). In addition, the complete camera model is provided for each image in the formm = K [R | −Rt] X, including all associated parameters (HV40_YYYY_calibrated_camera_parameters.txt). The dataset enables evaluation, validation, and comparison of image-based monitoring and change detection approaches under challenging high-alpine environmental conditions. Information on the Hochvogel test site, the long-term monitoring setup, and related research activities can be found on the AlpSenseRely project website. Additional methodological details are provided in the puplication “Feature-based multi-epoch rock slope monitoring using images and terrestrial laser scans” (Lucks & Holst, 2026). Citation of this work is kindly requested when using this dataset .