Self-Supervised Multi-Scale Transformer with Attention-Guided Fusion for Efficient Crack Detection

2025-10-11

AI Summary

Road crack detection typically relies on costly pixel-level annotations, hindering scalable deployment. To address this, we propose Crack-Segmenter—the first fully self-supervised pixel-level segmentation framework for pavement cracks. Our method innovatively integrates scale-adaptive embeddings, orientation-aware Transformers, and attention-guided multi-scale feature fusion, while incorporating pixel-wise contrastive learning to enable high-fidelity representation learning without any manual labels. Extensive evaluation across 10 public benchmarks demonstrates that Crack-Segmenter consistently outperforms 13 state-of-the-art supervised methods in key metrics—including mIoU, Dice coefficient, XOR error, and Hausdorff distance—achieving, for the first time, high-accuracy crack segmentation without any human annotation. This breakthrough significantly enhances the scalability and practicality of intelligent infrastructure monitoring at scale.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Unsupervised & Self-Supervised Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Pavement crack detection has long depended on costly and time-intensive pixel-level annotations, which limit its scalability for large-scale infrastructure monitoring. To overcome this barrier, this paper examines the feasibility of achieving effective pixel-level crack segmentation entirely without manual annotations. Building on this objective, a fully self-supervised framework, Crack-Segmenter, is developed, integrating three complementary modules: the Scale-Adaptive Embedder (SAE) for robust multi-scale feature extraction, the Directional Attention Transformer (DAT) for maintaining linear crack continuity, and the Attention-Guided Fusion (AGF) module for adaptive feature integration. Through evaluations on ten public datasets, Crack-Segmenter consistently outperforms 13 state-of-the-art supervised methods across all major metrics, including mean Intersection over Union (mIoU), Dice score, XOR, and Hausdorff Distance (HD). These findings demonstrate that annotation-free crack detection is not only feasible but also superior, enabling transportation agencies and infrastructure managers to conduct scalable and cost-effective monitoring. This work advances self-supervised learning and motivates pavement cracks detection research.

Problem

• Achieving pixel-level crack segmentation without manual annotations
• Overcoming costly pixel-level annotation limitations in crack detection
• Developing self-supervised framework for scalable infrastructure monitoring

Innovation

• Self-supervised framework eliminates manual annotation needs
• Multi-scale transformer maintains crack continuity with attention
• Attention-guided fusion module adaptively integrates features