融合注意力机制的安全帽检测算法

A Safety Helmet Detection Algorithm Integrating Attention Mechanism

针对工业施工场所背景复杂导致安全帽检测精度低及漏检等问题,提出一种融合注意力机制的安全帽检测算法。该算法在YOLOv5s网络模型的基础上,在主干网络中加入坐标注意力模块,使得网络可以有效关注目标信息的特征,提升远距离目标的检测能力。在网络训练过程中优化损失函数,将原有的CIoU损失函数更换为EIoU损失函数,优化了目标边界框回归的收敛速度,可以生成定位更精准的边界框,提高了模型检测精度。实验结果表明,改进后的算法平均精度达到94.5%,相较于原始模型提高了1.9个百分点,相较于YOLOv3算法提高了12.3个百分点。提出的算法有效地改善了原算法中安全帽漏检、误检的情况,同时提高了检测精度。

Aiming at solving the problems of low detection accuracy and missed detection caused by the complex background in industrial construction sites,a safety helmet detection algorithm integrating attention mechanism is proposed.Based on the YOLOv5s network model,the coordinate attention module is added to the backbone network so that the network can effectively focus on the characteristics of target information and improve the detection ability of long distance targets.The loss function is optimized by replacing the original CIoU loss function with EIoU loss function in the process of network training,and the convergence speed of target boundary box regression is optimized,which can generate boundary boxes with more accurate positioning and improve the accuracy of model detection.According to the experimental results,the average accuracy of the improved algorithm reaches 94.5%,which is 1.9 percentage points higher than the original model accuracy.Compared to the YOLOv3 algorithm,the accuracy has improved by 12.3 percentage points.The proposed algorithm can effectively improve the missed and false detection of safety helmets in the original algorithm,and also enhance the detection accuracy.