基于CBAM注意力机制的YOLOv5目标检测算法在电力业务证件识别上的应用

The Application of YOLOv5 Object Detection Algorithm Based on CBAM Attention Mechanism in the Identification of Credentials for Power Business

针对在电力业务领域传统的人工方式进行证件识别存在的效率低、识别时间长和可靠性低等问题,提出了一种用于电力业务工作中证件识别的改进YOLOv5模型。引入CBAM (Convolutional Block At-tention Module)算法提高特征提取性能,解决在图像分辨率低、光线暗等场景下识别率低的问题。通过对改进前后算法模型性能的对比分析,验证了该方法的优越性。实验结果表明,与原有的YOLOv5检测算法相比,所提方法在检测速度上能够满足实际检测的需要,且检测精度更优,检测时间为0.056 s,检测平均准确度均值为95.40%,提高了9个百分点。

An improved YOLOv5 model for certificate recognition in the power industry is proposed to address the problems of low efficiency, long recognition time, and low reliability of traditional manual methods. The CBAM (Convolutional Block Attention Module) algorithm is introduced to enhance feature extraction performance and solve the problem of low recognition rate in scenarios with low image resolution and dark lighting. Through comparative analysis of the performance of the algorithm model before and after improvement, the superiority of this method is verified. The experimental results show that compared with the original YOLOv5 detection algorithm, the proposed method can meet the actual detection needs in terms of detection speed, and has better detection accuracy. The detection time is 0.056 seconds, and the mean average precision of detection is 95.40%, which is increased by 9 percent.