Multi-Task Crack Foundation Model for Engineering-Reliable Crack Representation and Topology Preservation in Civil Infrastructure

2026-06-03

AI Summary

Existing crack segmentation methods often suffer from fragmented predictions and missed fine branches under domain shift, while also lacking reliable uncertainty estimates. To address these limitations, this work proposes CrackGeoFM, a multi-task foundational model for crack analysis that leverages a frozen vision foundation backbone augmented with three key components: a Frequency-guided Crack Enhancement Module (FCEM), a Crack-domain Feature Adaptation Module (CFAM), and a Structure-aware Multi-task Decoder (SMTD). This unified framework jointly performs pixel-level segmentation, skeleton reconstruction, and uncertainty calibration. By integrating frequency enhancement, domain adaptation, and structure-aware multi-task learning into a foundational model architecture—novel in the crack analysis domain—the method achieves state-of-the-art performance across 20 datasets, significantly improving topological completeness and uncertainty reliability, and enabling effective few-shot transfer with as few as five annotated images.

Abstract

Reliable crack assessment requires not only accurate pixel-level masks but also connected crack geometry and confidence estimates that remain stable under domain shift. However, existing segmentation models can achieve high overlap scores while fragmenting cracks, missing fine branches, and providing no calibrated uncertainty. To address this gap, this paper proposes CrackGeoFM, a multi-task framework that combines a frozen visual foundation backbone with crack-specific adaptation for mask prediction, skeleton reconstruction, and uncertainty estimation. The framework integrates a Frequency-Guided Crack Enhancement Module (FCEM) to enhance high-frequency crack cues, a Crack-Domain Feature Adaptation Module (CFAM) to adapt frozen backbone features to crack-domain patterns, and a Structure-Aware Multi-Task Decoder (SMTD) to jointly decode masks, skeletons, and uncertainty. Across 20 crack datasets, CrackGeoFM achieves state-of-the-art segmentation, improved topology preservation, calibrated uncertainty, and effective few-shot adaptation with only five labeled images. These results support reliable, generalizable, and engineering-oriented crack analysis for infrastructure assessment.

Problem

• crack segmentation
• topology preservation
• uncertainty estimation
• domain shift
• civil infrastructure

Innovation

• multi-task learning
• foundation model
• topology preservation
• uncertainty estimation
• few-shot adaptation