基于改进YOLOv5s的指针式水表读数检测

针对光照不均匀和水表表盘雾化的指针式水表在读数检测时出现漏检、误检等问题,本文提出一种基于改进YOLOv5s的指针式水表读数检测方法。首先,采用Mosaic,Mixup等数据增强方法,提高模型的泛化能力;其次,引入BiFPN模块(Bilateral Feature Pyramid Network)实现更高层次的特征融合,使得水表图像的深层特征图和浅层特征图充分融合,提高网络的表达能力;然后,嵌入CBAM注意力机制模块（Convolutional Block Attention Module）,在通道和空间双重维度上强化指针式水表子表盘示数特征;最后将CIoU-Loss函数(Complete Intersection over Union Loss)替换为SIoU_Loss,提升边界框的回归精度。改进算法的mAP@0.5达到97.8%,比YOLOv5s原始网络提升了3.2%。实验结果表明：该算法能有效提高指针式水表的读数检测精度。

Reading Detection of Pointer Water Meter Based on Improved YOLOv5s

This article proposes a pointer water meter reading detection method based on improved YOLOv5s to address issues such as missed or incorrect readings in pointer water meters with uneven lighting and atomized water meter dials. Firstly, data augmentation methods such as Mosaic and Mixup are used to improve the generalization ability of the model; Secondly, the BiFPN module (Bilateral Feature Pyramid Network) is introduced to achieve higher-level feature fusion, enabling the full fusion of deep and shallow feature maps of water meter images, thereby improving the network''s expression ability; Then, embed the CBAM Attention Mechanism Module (Convolutional Block Attention Module) to enhance the sub dial indication features of the pointer water meter in both channel and spatial dimensions; Finally, replace the CIoU Loss function (Complete Intersection over Union Loss) with SIoU_ Loss, improve the regression accuracy of bounding boxes. Improved algorithm mAP@0.5 Reached 97.8%, an improvement of 3.2% compared to the YOLOv5s original network. The experimental results show that this algorithm can effectively improve the reading detection accuracy of pointer water meters.