# HR-Mamba: Building Footprint Segmentation with Geometry-Driven Boundary Regularization

## TL;DR

HR-Mamba is a new high-resolution network for building footprint segmentation that improves accuracy in dense urban areas by combining boundary regularization and global modeling.

## Contribution

HR-Mamba introduces a novel architecture with geometry-driven boundary regularization and a Mamba-based global branch for better building extraction.

## Key findings

- HR-Mamba improves F1-score by 2.98% compared to HRNet in dense urban imagery.
- The model enhances detail fidelity and global consistency in building footprint segmentation.
- A composite loss and adaptive post-processing pipeline contribute to smoother, engineering-ready outlines.

## Abstract

Building extraction underpins land-use assessment, urban planning, and disaster mitigation, yet dense urban scenes still cause missed small objects, target adhesion, and ragged contours. We present High-Resolution-Mamba (HR-Mamba), a high-resolution semantic segmentation network that augments a High-Resolution Network (HRNet) parallel backbone with edge-aware and sequence-state modeling. A Canny-enhanced, median-filtered stem stabilizes boundaries under noise; Involution-based residual blocks capture position-specific local geometry; and a Mamba-based State Space Models (Mamba-SSM) global branch captures cross-scale long-range dependencies with linear complexity. Training uses a composite loss of binary cross entropy (BCE), Dice loss, and Boundary loss, with weights selected by joint grid search. We further design a feature-driven adaptive post-processing pipeline that includes geometric feature analysis, multi-strategy simplification, multi-directional regularization, and topological consistency verification to produce regular, smooth, engineering-ready building outlines. On dense urban imagery, HR-Mamba improves F1-score from 80.95% to 83.93%, an absolute increase of 2.98% relative to HRNet. We conclude that HR-Mamba jointly enhances detail fidelity and global consistency and offers a generalizable route for high-resolution building extraction in remote sensing.