synth-dacl: Does Synthetic Defect Data Enhance Segmentation Accuracy and Robustness for Real-World Bridge Inspections?

2025-06-17

AI Summary

Pixel-level defect segmentation in intelligent bridge inspection suffers from low accuracy and poor robustness, exacerbated by severe fine-grained class imbalance (e.g., cracks, voids) in the DACL10k dataset. Method: We propose the first synthetic defect data generation framework tailored to real-world bridge inspection scenarios. It leverages controllable concrete texture modeling and source-to-target distribution alignment to construct three synthetic expansion datasets (Synth-DACL). Contribution/Results: Integrated with semantic segmentation models (e.g., U-Net) and evaluated under 15 image degradation types, Synth-DACL boosts mIoU, F1-score, recall, and precision by 2% on average. It significantly enhances model generalization across diverse corruptions while offering a reproducible, scene-adaptive data synthesis paradigm for few-shot fine-grained defect segmentation.

Technology Category

• Computer Vision: Computational Photography, Image & Video Synthesis
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Machine Learning: Deep Generative Models & Autoencoders

Application Category

• Web Mining and Content Analysis: Large pretrained models with web data
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Adequate bridge inspection is increasingly challenging in many countries due to growing ailing stocks, compounded with a lack of staff and financial resources. Automating the key task of visual bridge inspection, classification of defects and building components on pixel level, improves efficiency, increases accuracy and enhances safety in the inspection process and resulting building assessment. Models overtaking this task must cope with an assortment of real-world conditions. They must be robust to variations in image quality, as well as background texture, as defects often appear on surfaces of diverse texture and degree of weathering. dacl10k is the largest and most diverse dataset for real-world concrete bridge inspections. However, the dataset exhibits class imbalance, which leads to notably poor model performance particularly when segmenting fine-grained classes such as cracks and cavities. This work introduces"synth-dacl", a compilation of three novel dataset extensions based on synthetic concrete textures. These extensions are designed to balance class distribution in dacl10k and enhance model performance, especially for crack and cavity segmentation. When incorporating the synth-dacl extensions, we observe substantial improvements in model robustness across 15 perturbed test sets. Notably, on the perturbed test set, a model trained on dacl10k combined with all synthetic extensions achieves a 2% increase in mean IoU, F1 score, Recall, and Precision compared to the same model trained solely on dacl10k.

Problem

• Enhancing bridge defect segmentation accuracy using synthetic data
• Addressing class imbalance in real-world bridge inspection datasets
• Improving model robustness to image quality and texture variations

Innovation

• Uses synthetic data to balance class distribution
• Enhances model robustness with synthetic textures
• Improves segmentation accuracy for fine-grained defects