基于YOLOv8-NFMC的带钢表面缺陷检测算法

Surface defect detection algorithm for strip steel based on YOLOv8-NFMC

针对YOLOv8算法在应用于带钢表面缺陷检测时存在漏检和错检等问题,提出了一种改进YOLOv8算法。针对数据集中的小目标的标签,在原损失CIOU的基础上面加入标准化高斯瓦瑟斯坦距离(normalized Gaussian Wasserstein distance,NWD),提升模型对小目标缺陷的检测能力;采用聚焦调制(focal modulation)替换YOLOv8模型的空间池化金字塔(spatial pyramid pooling-fast,SPPF),在轻量化的同时,提高多尺度特征的表达能力;采用移动翻转瓶颈卷积(mobile inverted bottleneck conv,MBConv)替换C2f中的Conv构建新模块C2f-MB,同时使用C2f-MB替换原有的C2f模块,增强特征表达能力和多尺度特征融合能力;在主干部分加入卷积块注意力机制(convolutional block attention module,CBAM)来抑制背景干扰,能更好捕获全局信息,提升了主干部分的特征提取能力。实验结果表明,改进后的YOLOv8算法在计算量下降的同时,mAP@0.5提高了3%,对漏检和错检等问题有明显改善。

Aiming at the problems of leakage and wrong detection when the YOLOv8 algorithm is applied to the surface defect detection of strip steel,an improved YOLOv8 algorithm is proposed.For the labels of small targets in the dataset,normalized Gaussian Wasserstein distance(NWD)is added on top of the original lossy CIOU,which improves the model's ability to detect defects of small targets.Focal Modulation is used to replace the spatial pooling pyramid of the YOLOv8 model,which improves the expression ability of multi-scale features while lightweighting.Mobile inverted bottleneck Conv(MBConv)is used to replace the Conv in C2f to construct a new module C2f-MB,and at the same time replace the original C2f-MB with C2f-MB.MB to replace the original C2f module with C2f-MB,which enhances the feature expression ability and multi-scale feature fusion ability.the convolutional block attention module(CBAM)is added in the backbone part to suppress the background interference,which can better capture the global information and improve the feature extraction ability of the backbone part.Experiment results show that the improved YOLOv8 algorithm improves mAP@0.5 by 3%while decreasing the computation amount,which significantly improves the problems of missed detection and wrong detection.