基于YOLOv8网络模型的电梯内电动自行车检测方法  

A Method for Electric Bicycle Detection in Elevator Based on YOLOv8 Network Model

作　　者：王栩漓 王玲芝 吴英凡 WANG Xuli;WANG Lingzhi;WU Yingfan(School of Automation,Xi′an University of Posts and Telecommunications,Xi′an 710121,China)

机构地区：[1]西安邮电大学自动化学院,陕西西安710121

出　　处：《成组技术与生产现代化》2024年第3期8-12,共5页Group Technology & Production Modernization

基　　金：陕西省自然科学基础研究计划资助项目(2023-JC-QN-0515);西安邮电大学研究生创新基金资助项目(CXJJYL2023066)。

摘　　要：针对电动自行车进入电梯,入户充电而危及人身安全的问题,提出了一种基于深度学习的电梯内电动自行车自动检测方法。以YOLOv8作为核心检测的网络模型,通过构建数据集进行了训练和测试。结果表明,YOLOv8网络模型能够对电梯中违规进入的电动自行车进行精准识别,即使在电动自行车被部分遮挡的情况下也能保持较高的识别率,证明了YOLOv8网络模型在不同场景下应用的有效性和稳定性。由对比实验可知,YOLOv8网络模型相对YOLOv5网络模型来说,不仅可提高安全管理的效率,而且能够及时预警,显著提升电梯乘客的安全。与传统人工检测和视频监控方法相比,提出的电梯内电动自行车检测方法能够实时处理图像并精确识别目标,具有较高的实用价值和广泛的应用前景。The entry of electric bicycles into elevator for charging poses significant safety risks to residents and may lead to fire hazards and life safety issues.To solve this problem,an automatic detection method for electric bicycles inside elevators was put forward in this study based on deep learning techniques.The YOLOv8 network model was used as the core detection model and was trained and tested using a constructed dataset.Experimental results showed that the model could accurately identify electric bicycles entering elevators even when partially obstructed,highlighting the effectiveness and robustness of the YOLOv8 network model in various scenarios.Furthermore,comparative experiments with the YOLOv5 network model confirmed the superiority of the YOLOv8 network model.This approach not only enhanced safety management efficiency,but also provided timely warnings to significantly improve elevater passenger safety.In comparison to traditional manual detection method and video surveillance system,the proposed electric bicycle detection method offered real-time image processing capabilities and precise target identification with broad practical value and application prospect.

关 键 词：电梯安全 电动自行车检测 YOLOv8 目标检测 网络模型 

分 类 号：TP23[自动化与计算机技术—检测技术与自动化装置]