GBA-UBF : A Large-Scale and Fine-Grained Building Function Classification Dataset in the Greater Bay Area

TL;DR

This paper presents GBA-UBF, a large-scale, building-level function dataset for six core cities in the Guangdong–Hong Kong–Macao Greater Bay Area, containing five functional classes and covering nearly four million buildings. It introduces the Multi-level Building Function Optimization (ML-BFO) pipeline that fuses POI data with building footprints through candidate label generation, neighborhood-based refinement, and high-level POI-driven corrections, coupled with the Building Function Matching Index (BFMI) to evaluate spatial and semantic alignment with POI heatmaps. The results show superior granularity and accuracy over parcel-based baselines (e.g., EULUC-China 2.0), validated both quantitatively and via field checks, with demonstrated applicability to urban planning, risk assessment, and smart-city analytics. By providing a reproducible workflow, open data, and a robust evaluation metric, the work advances fine-grained urban analytics and offers a practical resource for city governance and digital twin applications in a densely populated, morphologically diverse megaregion.

Abstract

Rapid urbanization in the Guangdong-Hong Kong-Macao Greater Bay Area (GBA) has created urgent demand for high-resolution, building-level functional data to support sustainable spatial planning. Existing land use datasets suffer from coarse granularity and difficulty in capturing intra-block heterogeneity. To this end, we present the Greater Bay Area Urban Building Function Dataset (GBA-UBF), a large-scale, fine-grained dataset that assigns one of five functional categories to nearly four million buildings across six core GBA cities. We proposed a Multi-level Building Function Optimization (ML-BFO) method by integrating Points of Interest (POI) records and building footprints through a three-stage pipeline: (1) candidate label generation using spatial overlay with proximity weighting, (2) iterative refinement based on neighborhood label autocorrelation, and (3) function-related correction informed by High-level POI buffers. To quantitatively validate results, we design the Building Function Matching Index (BFMI), which jointly measures categorical consistency and distributional similarity against POI-derived probability heatmaps. Comparative experiments demonstrate that GBA-UBF achieves significantly higher accuracy, with a BMFI of 0.58. This value markedly exceeds that of the baseline dataset and exhibits superior alignment with urban activity patterns. Field validation further confirms the dataset's semantic reliability and practical interpretability. The GBA-UBF dataset establishes a reproducible framework for building-level functional classification, bridging the gap between coarse land use maps and fine-grained urban analytics. The dataset is accessible at https://github.com/chenchs0629/GBA-UBF, and the data will undergo continuous improvement and updates based on feedback from the community.

Figures (9)

• Figure 1: Geographical Location of the GBA and Study Area.
• Figure 2: Location and Data Overview of the Study Area within the Greater Bay Area. Different colors represent different cities, and for each city we show both the distribution of POI and representative building footprints.
• Figure 3: Overall of dataset generation framework. We use a reproducible 3-stage pipeline: (a) Candidate Function Label Generation; (b) Iterative Label Distribution Refinement; and (c) Function Related Label Correction.
• Figure 4: Construction of the Building Function Matching Index (BFMI). POI records are converted to class-specific KDE heatmaps and compared with the final building-level labels. For each building we compute (1) Top-1 Consistency with the dominant POI class and (2) Cosine Similarity between the one-hot label and the POI-derived probability vector, to generate the BFMI score map.
• Figure 5: Distribution of Building Functional Categories in the Central Urban Areas of Six GBA Cities.