Muemmel/solar-pv-detection-brandenburg-dataset

Solar PV Ground Truth Dataset — Brandenburg Orthophotos

Hand-corrected ground-truth masks for photovoltaic detection on 20 cm GSD aerial orthophotos (1 km × 1 km tiles, 4-band RGBI) of Brandenburg, Germany. Companion to the model repository solar-pv-segmentation-brandenburg.

Contents

Folder	Contents
gt_masks_selected/	1432 hand-corrected patch masks (256×256 px, uint8, 0=background / 1=PV), split into training/validation/testing
gt_masks_full_tiles/	50 mosaicked full-tile masks (5000×5000 px), same encoding
metadata/tiles.csv	50 tiles: tile ID (dop_x-y), split, aerial recording date
metadata/plants_mastr.csv	38 PV plants (50 plant×tile pairs) linked to their official MaStR unit IDs, mounting type, capacity, commissioning date
dop_tiles/	the 50 corresponding DOP tiles (RGBI, zstd-compressed GeoTIFF, 0.2 m GSD), organized by split — byte-identical to the LGB source delivery, so tile selection and flight dates are verifiable

Splits: 30 training / 11 validation / 9 testing tiles; patches were sampled from the 19×19 inner grid of each tile (only patches containing ≥ 1.6 m² of PV).

Source data and license

• Imagery: © GeoBasis-DE/LGB (recordings 2021–2024), dl-de/by-2-0, Daten geändert (Datenlizenz Deutschland — Namensnennung 2.0; compatible with CC BY 4.0).
• Plant metadata: Marktstammdatenregister (MaStR), Bundesnetzagentur — public bulk download (2025-01-12), minimal subset published here (unit IDs, mounting type, capacity, dates — no operator or address data).

This dataset is a derivative work of the LGB imagery; the attribution above is mandatory for any redistribution. The masks themselves are released under CC BY 4.0.

Class encoding

Masks are single-band uint8 GeoTIFFs aligned to the DOP grid: 0 = background, 1 = photovoltaic module area.

Method

Annotations started from MaStR plant locations (≥ 30 kWp, recorded 2021–2024), selected with a fixed seed to match the Brandenburg roof/ground ratio, then hand-labeled and hand-corrected on the 0.2 m orthophotos. Inner patches without PV (≥ 1.6 m²) are not included in the selected subset; the full-tile masks cover complete 1 km tiles including edge patches.

Intended use

• Training/evaluating PV segmentation models on German orthophotos.
• Validation of energy-register data quality (MaStR) via remote sensing.
• The splits are spatially disjoint (tile-level); do not re-shuffle.

Citation

Master thesis: Validation of Photovoltaic Data in the German Core Energy Market Register via Aerial Image Segmentation, Hochschule Harz (2026). Model checkpoints: https://huggingface.co/Muemmel/solar-pv-segmentation-brandenburg