Facade Segmentation for Solar Photovoltaic Suitability

TL;DR

Addressing the challenge of designing BIPV facades when roof space is insufficient, the paper presents a pipeline that from semantic facade segmentation derives PV-suitability masks and practical panel layouts, then estimates energy yield. It fine-tunes SegFormer-B5 on the CMP Facades dataset to produce 13-class wall-component predictions and converts them into PV masks via clearance-aware filtering, followed by quadrant-based panel placement within the largest contained rectangle and PVGIS energy yield calculations. On CMP, the model achieves a dataset-level mIoU of 0.52 with macro F1 ≈ 0.68, while application to 373 LSAA facades shows a large gap between theoretical (≈39% of wall area) and practical installable PV (≈4–5%), due to occlusions and complex ornamentation. The results highlight the value of translating surface segmentation into design-ready layouts and energy forecasts, enabling scalable urban planning and inclusion in solar cadastres. Limitations include occlusions, pitched roofs, and single-rectangle packing; future work proposes occlusion removal, multi-region packing, 3D shading simulations, and design alternatives beyond maximum coverage.

Abstract

Building integrated photovoltaic (BIPV) facades represent a promising pathway towards urban decarbonization, especially where roof areas are insufficient and ground-mounted arrays are infeasible. Although machine learning-based approaches to support photovoltaic (PV) planning on rooftops are well researched, automated approaches for facades still remain scarce and oversimplified. This paper therefore presents a pipeline that integrates detailed information on the architectural composition of the facade to automatically identify suitable surfaces for PV application and estimate the solar energy potential. The pipeline fine-tunes SegFormer-B5 on the CMP Facades dataset and converts semantic predictions into facade-level PV suitability masks and PV panel layouts considering module sizes and clearances. Applied to a dataset of 373 facades with known dimensions from ten cities, the results show that installable BIPV potential is significantly lower than theoretical potential, thus providing valuable insights for reliable urban energy planning. With the growing availability of facade imagery, the proposed pipeline can be scaled to support BIPV planning in cities worldwide.

Figures (2)

• Figure 1: Example applications (with panel size L) from the LSAA subset (N = 373).
• Figure 2: Failure modes due to occlusions (a), segmentation accuracy (b), complexity of facade (c).