基于YOLOv8的吸烟行为检测研究--毕业论文（Python、深度学习、YoLov8/YOLOv5）

摘  要

随着公共健康问题日益受到关注，吸烟行为的监测成为了提升健康管理和控制吸烟行为的重要手段。传统的吸烟行为监测方法依赖人工检查，工作量大且效率低。随着计算机视觉和深度学习技术的发展，基于视觉的吸烟行为检测逐渐成为一种新兴的研究方向。YOLO（You Only Look Once）系列目标检测算法以其高效、快速的特点在实时物体检测中得到广泛应用，因此，基于YOLO的吸烟行为检测在准确性和实时性上具有显著优势。本文针对YOLOv8算法在吸烟行为检测中的应用进行了研究与实验，结合YOLOv7、YOLOv6和YOLOv5算法，进行性能对比分析，以期为吸烟行为的实时监控提供一种高效的解决方案。

本文的主要研究内容包括：首先，介绍了国内外吸烟行为检测的研究现状，分析了现有方法的优势与不足；其次，提出了一种基于YOLOv8的吸烟行为检测系统，结合YOLOv7、YOLOv6和YOLOv5算法对比，分析不同算法在检测性能、精度和速度方面的差异；再次，探讨了数据集的处理方法，并详细阐述了算法原理与模型训练的过程；此外，设计了一个基于Streamlit的交互式Web应用界面，实现了图像、视频和实时摄像头的吸烟行为检测，支持不同YOLO训练模型的上传与推理预测，用户可根据需要修改检测设置，进一步提升了系统的灵活性与实用性。

该研究的主要优点在于，基于YOLOv8的吸烟行为检测系统不仅能实现高精度、高效率的实时检测，还具备灵活的界面设计，支持多种训练模型与推理预测，能够在不同应用场景中提供定制化的服务。

关键词：YOLOv8；吸烟行为检测；深度学习；Streamlit；目标检测

Abstract

With increasing public health concerns, monitoring smoking behaviors has become an important means to improve health management and control smoking behaviors. Traditional methods of monitoring smoking behavior rely on manual inspection, with heavy workload and low efficiency. With the development of computer vision and deep learning technologies, vision-based smoking behavior detection has gradually become an emerging research direction. YOLO (You Only Look Once) series target detection algorithm is widely used in real-time object detection with its high efficiency and fast characteristics, so YOLO-based smoking behavior detection has significant advantages in accuracy and real-time. In this paper, the application of YOLOv8 algorithm in the detection of smoking behavior was studied, combining YOLOv7, YOLOv6 and YOLOv5 algorithms, in order to provide an efficient solution for the real-time monitoring of smoking behavior.

The main research contents of this paper include: First, Introduced the research status of smoking behavior detection at home and abroad, The advantages and disadvantages of the existing methods are analyzed; next, Proposed a YOLOv8-based detection system for smoking behavior, Combining the comparison of YOLOv7, YOLOv6 and YOLOv5 algorithms, Analyze the differences in the detection performance, accuracy and speed of different algorithms; once more, Explore the processing methods of the dataset, The process of algorithm principle and model training is described in detail; besides, Designed an interactive Web application interface based on Streamlit, Implement the smoking behavior detection of images, video and real-time cameras, Support the upload and inference prediction of different YOLO training models, Users can modify the detection settings as necessary, It further improves the flexibility and practicality of the system.

The main advantages of this study are that the YOLOv8-based smoking behavior detection system can not only achieve high precision and efficient real-time detection, but also has flexible interface design, support a variety of training models and reasoning prediction, and can provide customized services in different application scenarios.

Keywords: YOLOv8; smoking behavior detection; deep learning; Streamlit; target detection

目  录<