Automated metadata extraction for solar PV installations using satellite imagery and computer vision.
Panel-Segmentation provides tools to detect, locate, and characterize solar PV installations from satellite and aerial imagery. It is developed at the National Renewable Energy Laboratory (NREL).
Capabilities include:
- Satellite image retrieval — pull imagery for any lat/lon coordinate via the Google Maps Static API
- Solar panel detection — locate panels and retrieve their latitude, longitude, and address within a geographic bounding box (SOL-Searcher Pipeline)
- Storm damage assessment — detect and quantify hurricane or hail damage on solar installations from before/after satellite imagery
- NOAA MESH conversion — convert NOAA Maximum Estimated Size of Hail (MESH) grib2 files to KML or GeoJSON
- Tilt and azimuth estimation — estimate array tilt and azimuth from USGS LiDAR data
- Mounting configuration classification — identify roof-mount, ground-mount, carport, and other configurations using an instance segmentation model
- Site autotagging — geolocate tracker rows, fixed-tilt rows, and inverters at utility-scale ground-mount sites using a deep learning pipeline
Panel-Segmentation stores deep learning model weights via Git Large File Storage (LFS). Install Git LFS before cloning:
# macOS
brew install git-lfs
# Linux
sudo apt install git-lfs
# Then initialize
git lfs installgit clone https://github.com/NREL/Panel-Segmentation.git
cd Panel-Segmentation
pip install .All CV models depend on MMCV, which must be built from source:
pip install git+https://github.com/open-mmlab/mmcv.git@v2.1.0Note: MMCV installation may take several minutes.
Full documentation — including installation guides, API reference, pipeline tutorials, and example notebooks — is available at panel-segmentation.readthedocs.io.
Contributions are welcome. Please open an issue or pull request on GitHub.
If you use Panel-Segmentation in your research, please cite it using the following citation:
K. Perry and C. Campos, "Panel Segmentation: A Python Package for Automated Solar Array Metadata Extraction Using Satellite Imagery," in IEEE Journal of Photovoltaics, vol. 13, no. 2, pp. 208-212, March 2023, doi: 10.1109/JPHOTOV.2022.3230565.
Panel-Segmentation is released under the BSD 3-Clause License.