Automated pipeline generates realistic crack images for inspection data

An End-to-End Automated Pipeline for Controllable Crack Data Synthesis

Computer Vision and Pattern Recognition

Summary

Inspecting cracks with AI tools is hard because there aren’t many examples of cracks to learn from, and existing methods can’t easily control how cracks look or fit into scenes. The authors created a tool that automatically makes lots of detailed crack images by first drawing crack shapes using mathematical curves and then turning those into realistic crack pictures with AI. Their method can also add cracks into different backgrounds, improving how well AI can learn to find cracks. They tested their approach on popular crack image sets and showed it works better than older methods.

What this means in practice

Authors

Conghui Li, Muxin Pu, Chern Hong Lim, Weiyao Lin, Xin Wang

Abstract

Automated crack inspection increasingly relies on deep learning, yet its reliability is limited by scarce and weakly controllable defect data. Existing generative augmentation methods often treat crack synthesis as a generic image-generation task, offering insufficient control over morphology, boundary fidelity, and scene context. This paper proposes an end-to-end automated pipeline for controllable crack data synthesis that formalizes crack geometry and inspection context into reusable computational constraints. First, procedurally sampled Bézier-curve skeletons are translated into realistic crack masks using a GAN, enabling scalable generation of diverse crack morphologies without manual mask design. Second, a dual-ControlNet diffusion framework disentangles appearance guidance from geometric guidance, with an edge-based branch enforcing strict boundary consistency. The framework supports both background-free synthesis and context-aware inpainting. Experiments on CRACK500 and CrackTree200 show consistent gains over existing augmentation baselines, demonstrating a scalable engineering informatics workflow for automated crack-inspection data generation.