An open-source, deep learning-based pipeline that maps rooftop photovoltaic installations from aerial imagery, at national scale — and the research foundation Solar Dashboard is built on.
Rooftop PV systems mapped by DeepPVMapper across France — the darker, the higher the installed capacity. Explore the project →
The pipeline
DeepPVMapper is a two-stage deep learning pipeline, inspired by 3D-PV-Locator (Meyer et al., 2022). A classification model (Inception v3) first flags candidate image patches on aerial imagery; these patches are then segmented with DeepLab v3 to extract precise rooftop PV polygons. Detections are cross-referenced with the BD TOPO® building database to keep only rooftop-mounted systems and merge detections belonging to the same roof, then characterized with PyPVRoof (surface, tilt, orientation, installed capacity) to produce a geolocated PV registry. The models are trained on BD-PV, an annotated dataset built on IGN's BD ORTHO® aerial imagery.
From aerial imagery to geolocated PV registry — explore the pipeline in detail →
Auditing grid-connection registries
Because DeepPVMapper's detections are independent of administrative records, they can be used to audit official grid-connection registries (RTE/RNI) rather than the other way around. After correcting for the pipeline's measured precision and recall by département, the resulting estimates reveal a systematic underestimation of rooftop PV in these registries, particularly in rural areas. Read the full analysis →.
Click here 
to download the minimum data (images, model weights, additional data) to replicate the example provided in the
GitHub repository, or here 
to download the full registry of 520k+ detected PV systems.
References:
- Kasmi, G., Dubus, L., Blanc, P., & Saint-Drenan, Y. M. (2022). Towards unsupervised assessment with open-source data of the accuracy of deep learning-based distributed PV mapping. arXiv preprint arXiv:2207.07466.
- Kasmi, G. (2024). Enhancing the Reliability of Deep Learning Models to Improve the Observability of French Rooftop Photovoltaic Installations. PhD thesis, Université Paris sciences et lettres.