摘要
针对智慧电梯安全感知系统要求检测算法具备轻量化以及快速推理的特点,提出基于改进YOLOv4的电梯内电动车检测算法。首先,采用轻量化网络GhostNet作为特征提取网络,减少模型的计算量及参数量;其次,引入CBAM注意力模块,提高算法的检测精度;最后,将FocalLoss机制应用到模型置信度损失中,平衡正负样本。实验结果表明,改进后的YOLO-GCF在电梯内电动车数据集上检测精度为90.14%,参数量减少82.8%,检测速度提升8.8帧/秒,做到了轻量化及快速推理。
Aiming at the intelligent elevator safety awareness system,which requires the detection algorithm to be lightweight and fast reasoning,a detection algorithm for electric bikes in elevator based on improved YOLOv4 is proposed.Firstly,the lightweight network GhostNet is used as the feature extraction network to reduce the calculation and parameter amount of the model.Secondly,CBAM attention module is introduced to improve the detection accuracy of the algorithm.Finally,the FocalLoss mechanism is applied to the model confidence loss to balance positive and negative samples.The experimental results on the data set of electric bikes in elevator show that the improved YOLO-GCF has a detection accuracy of 90.14%,a parameter reduction of 82.8%,and a detection speed increase of 8.8 frames per second,achieving lightweight and fast reasoning.
作者
杨献瑜
Yang Xianyu(School of Computer Science and Technology,Zhejiang Sci-Tech University,Hangzhou,Zhejiang 310018,China)
出处
《计算机时代》
2023年第10期54-58,共5页
Computer Era
