结合半监督和分割的墙体裂缝检测方法

A Wall Crack Detection Method Combining Semi-Supervision and Segmentation

为了检测墙体出现的裂缝,本文设计了一种基于半监督训练和曲线分割算法的裂缝检测算法,以保证建筑物和人的安全。该方法创新性地结合了半监督的训练方法,使用少量的标签数据和大量的无标签的数据训练模型,并提出使用基于曲线分割算法以有效分割裂缝,有效提高了墙体裂缝检测的效率和性能,实现了对复杂场景下裂缝的检测,在测试数据集上检测任务取得了优异的效果(检测精度达到91.55%),具有较强的鲁棒性,可以适用于多种复杂环境。

In order to detect the cracks appearing in the wall,this paper designes a crack detection algorithm based on semi-supervised training and curve segmentation algorithm,ensuring the safety of buildings and people.The method innovatively combines a semi-supervised training method to achieve the use of a small amount of labelled data and a large amount of unlabelled data to train the model,and proposes the use of a curve-based segmentation algorithm in order to efficiently segment the cracks,which effectively improves the efficiency and performance of the wall crack detection,and realises the detection of cracks in complex scenarios,and achieves excellent results in the detection task on the test dataset(with a detection accuracy of 91.55%),has strong robustness,and can be applied to a variety of complex environments.