SGL-Mamba: Structure-Aware Global–Local Mamba for Crack Segmentation

Abstract

Highlights What are the main findings? This paper proposes the SGL-Mamba network, introducing an SGL-Mamba block that combines a parallel directional scanning mechanism with a local perception branch to effectively achieve joint modeling of global topological continuity and local crack details. A high–low frequency separation enhancement module is incorporated to adaptively suppress background noise, while a deformable large kernel attention decoder is integrated to dynamically capture and adapt to irregular crack orientations. What are the implications of the main findings? Extensive evaluations on Crack500, DeepCrack, and CrackMap datasets demonstrate that SGL-Mamba consistently outperforms existing state-of-the-art methods, exhibiting accurate target delineation quality, structural coherence, and strong robustness against complex backgrounds. By successfully balancing a lightweight network architecture with high computational efficiency, this method provides a highly promising and practical solution for automated, high-precision crack detection in routine infrastructure health monitoring and safety assessments. Abstract Accurate crack segmentation based on optical sensor imagery is of paramount importance for infrastructure health monitoring and disaster early warning. However, in complex natural scenes, cracks typically exhibit characteristics such as being thin and elongated, multi-branched, and having low contrast. Existing segmentation models struggle to balance low computational overhead with the simultaneous modeling of global topological continuity and the precise extraction of local details. To overcome the aforementioned limitations, we introduce a Structure-Aware Global–Local Mamba (SGL-Mamba). Firstly, the SGL-Mamba Block is designed in the encoding stage, which significantly enhances the joint modeling capacity for continuous topological structures and edge textures through the synergy of a parallel directional scanning mechanism and a local perception branch. Secondly, we design a High–Low Frequency Separation Enhancement (HLFSE) module to reconstruct the skip connections. This module leverages frequency decoupling to adaptively suppress high-frequency background noise and alleviate the semantic gap. Finally, in the decoding stage, the Deformable Large Kernel Attention (D-LKA) is integrated, utilizing a dynamic spatial receptive field to precisely adapt to irregular crack orientations. Extensive experiments on three public datasets (Crack500, DeepCrack, and CrackMap) demonstrate that SGL-Mamba outperforms other state-of-the-art (SOTA) methods, achieving an F1 score (F1) of 0.7982 ± 0.0038 and an mIoU of 0.8011 ± 0.0036 on the Crack500 dataset. While ensuring lightweight architecture and high computational efficiency, the proposed method provides an effective and practical solution for automatic crack detection.