TY -的A2 -李,杨AU -史,Dongmei AU - Tang宏宇PY - 2021 DA - 2021/04/15 TI -安全驾驶评价方法研究基于机器视觉和长期短期记忆网络SP - 9955079六世- 2021 AB -交通运输业的迅速发展带来了一些潜在的安全隐患。针对行车安全的问题,人工智能技术的应用在安全驾驶行为识别可以有效地降低事故率和经济损失。根据干扰信号的存在,如时空背景混合信号驱动监控视频序列,小目标,如人眼的识别精度较低。本文提出一种改进的dual-stream卷积网络意识到安全驾驶的行为。基于卷积神经网络(cnn),注意机制(AM)集成到一个长短期记忆(LSTM)神经网络结构,以及混合dual-stream AM-LSTM卷积网络通道设计。空间河道使用CNN方法来提取视频图像的空间特征值并使用池,而不是传统的金字塔池、正常化的尺度转换。时间河道使用单发multibox检测器(SSD)算法来计算相邻两帧的视频序列检测等小物件的脸和眼睛。然后,AM-LSTM用于熔断器和dual-stream分类信息。自建的驾驶行为是建立视频图像集。中华民国、准确率和损失函数FDDB数据库中进行实验,VOT100数据集,分别和自建的视频图像集。 Compared with CNN, SSD, IDT, and dual-stream recognition methods, the accuracy rate of this method can be improved by at least 1.4%, and the average absolute error in four video sequences can be improved by more than 2%. On the contrary, in the self-built image set, the recognition rate of doze reaches 68.3%, which is higher than other methods. The experimental results show that this method has good recognition accuracy and practical application value. SN - 2090-0147 UR - https://doi.org/10.1155/2021/9955079 DO - 10.1155/2021/9955079 JF - Journal of Electrical and Computer Engineering PB - Hindawi KW - ER -