基于YOLOv5的施工振动邻近区域房屋裂缝智能检测

Intelligent Detection of House Cracks in the Vicinity of Construction Vibrations Based on YOLOv5

施工振动可能导致邻近区域的房屋产生裂缝,传统的裂缝检测技术由于现场环境和裂缝形态等因素影响检测效率,为此将智能检测技术运用于房屋裂缝检测中。采用YOLOv5算法,构建包含3917张图片的混凝土裂缝数据集,经过1000个训练轮次后,模型精度和召回率均达到较高水平,同时损失函数减小至极小值,表明该模型用于裂缝检测的精确性。将训练的模型用于38所房屋的裂缝检测中,检测效果良好。

Construction vibration may cause cracks in adjacent buildings.Traditional crack detection techniques are affected by factors such as on-site environment and crack morphology,which affect detection efficiency.Therefore,intelligent detection technology is applied to crack detection in buildings.The YOLOv5 algorithm is adopted,and a concrete crack dataset containing 3917 images is constructed.After 1000 training epochs,the model achieves high precision and recall rates,while the loss function decreases to a small value,indicating the accuracy of the model for crack detection.The trained model is applied to crack detection in 38 houses in practical engineering projects,showing good detection performance.