GATA2Floor: Graph attention for floor counting in street-view facades

TL;DR

GATA2Floor introduces a graph attention model for accurate, interpretable building floor counting from street-view facades, leveraging relational reasoning and self-supervised features without requiring extensive labeled data.

Contribution

The paper presents GATA2Floor, a novel graph attention-based model that predicts building floor counts and assigns facade elements to floors with interpretability and robustness, even without labeled datasets.

Findings

Effective floor counting from street-view images.

Robustness to irregular building designs.

No need for extensive labeled datasets due to self-supervised approach.

Abstract

Automated analysis of building facades from street-level imagery has great potential for urban analytics, energy assessment, and emergency planning. However, it requires reasoning over spatially arranged elements rather than solely isolated detections. In this work, we model each facade as a graph over window/door detections with a vertical prior on edges. Additionally, we introduce GATA2Floor, a multi-head Graph Attention v2 (GATv2) based model that predicts the global floor count of a building and, via learnable cross-attention queries, softly assigns elements to latent floor slots, yielding interpretable outputs and robustness to irregular designs. To mitigate the lack of labeled datasets, we demonstrate that the proposed graph-based reasoning can be applied without annotations by leveraging a lightweight label-free proposal mechanism based on self-supervised features and vision-language scoring. Our approach demonstrates the value of graph-attention-based relational reasoning for facade understanding.