Beyond Segmentation: Structurally Informed Facade Parsing from Imperfect Images

TL;DR

This paper enhances facade parsing by integrating a lightweight alignment loss into YOLOv8, improving structural regularity and geometric coherence in architectural element detection from imperfect images.

Contribution

It introduces a novel alignment loss to YOLOv8 training that enforces grid consistency, improving structural coherence without changing the inference process.

Findings

Improved structural regularity in facade parsing results.

Corrected alignment errors due to perspective and occlusion.

Maintained detection accuracy while enhancing geometric coherence.

Abstract

Standard object detectors typically treat architectural elements independently, often resulting in facade parsings that lack the structural coherence required for downstream procedural reconstruction. We address this limitation by augmenting the YOLOv8 training objective with a custom lightweight alignment loss. This regularization encourages grid-consistent arrangements of bounding boxes during training, effectively injecting geometric priors without altering the standard inference pipeline. Experiments on the CMP dataset demonstrate that our method successfully improves structural regularity, correcting alignment errors caused by perspective and occlusion while maintaining a controllable trade-off with standard detection accuracy.