基于YOLOv5的墙体裂缝检测

Wall crack detection based on YOLOv5

墙体裂缝检测对于居民建筑的保养意义较大,传统人工检测难以适应当前快速精确的检测需求,由计算机进行识别的目标检测算法恰好能解决此问题。卷积神经网络是当下目标检测的主流核心网络,采用高效算法对目标图像的处理办法,使计算成本大幅降低,被广泛运用于目标检测识别中。本文通过对深度学习的目标检测算法研究,选用YOLOv5模型,对常见的墙体裂缝进行识别。实验结果表明,YOLOv5算法模型能够有效识别数据集中包含的裂缝信息,精度达到0.845,召回率达到0.929,均值平均精度(mean Average Precision,mAP)达到了0.825。

Wall crack detection is of great significance for the maintenance of residential buildings.Traditional manual detection is difficult to meet the current rapid and accurate detection needs.The target detection algorithm which is recognized by computer can solve this problem.Convolutional neural network is the mainstream core network of object detection.It uses efficient algorithms to process target images,which greatly reduces the cost of computation and is widely used in object detection and recognition.By studying the deep learning target detection algorithm,this paper selects YOLO v5 model to identify common wall cracks.The experimental results show that the YOLOv5 algorithm model can effectively identify the crack information contained in the data set,and the accuracy reaches 0.847,the recall rate reaches 0.929,and the mean Average Precision(mAP)reaches 0.825.