Research on Pedestrian Detection and Tracking Technology Bas
发布时间：2021-08-13 21:43
信息的载体可以是语音信号,图像或者是视频。相较于语音信号,图像或视频中包含的信息要远远大于语音信号。如何从图像或视频中提取出有效信息就成为了一个热门的研究话题。计算机视觉是机器学习的一个热门方向,它的研究目标是让计算机能够模拟人的大脑,对输入的图像和视频进行理解分析。计算机视觉的任务主要包括图像分类,目标识别,目标跟踪和语义分割。传统的计算机视觉算法手工提取图像特征,比如空间特征,颜色通道特征,频率分布直方图等。随着深度学习研究的深入和并行计算硬件的提升,卷积神经网络开始被广泛应用于计算机视觉算法中。图像和视频的一个主要的采集设备是摄像头,而摄像头主要的拍摄对象之一是行人。在实际应用中,利用计算机视觉进行行人的检测与跟踪,并在此基础上进行行人的行为统计和分析,可以给视频监控和人流管理规划提供重要的技术支持和数据信息基础。比如,在无人驾驶领域,行人检测算法可以检测出出现在车载摄像头拍摄区域的行人;无人驾驶行车控制系统可以利用检测得到的行人位置以及轨迹数据近一步进行分析并调整行驶状态和规划路线。人流密集场景中,利用行人计数算法可以统计分析人流数量,并利用这些数据进行安全预警和人流疏导。所以...
【文章来源】：华中师范大学湖北省 211工程院校 教育部直属院校
【文章页数】：80 页
【学位级别】：硕士
【文章目录】：
Abstract
Chapter 1 Introduction
1.1 Research Background and Significance
1.2 Research Status
1.3 Research Difficulties
1.4 Main Content and Thesis Structure
1.5 Conclusion
Chapter 2 Basic Principle
2.1 Overview
2.2 Principles Involved in Pedestrian Detection
2.2.1 YOLO v3
2.2.2 YOLO v3 Tiny
2.3 Principles Involved in Pedestrian Tracking
2.3.1 Intersection Over Union
2.3.2 Hungarian Algorithm
2.3.3 Kalman Filter
2.4 Related Concepts of Pedestrian Tracking Based on MOT
2.5 Conclusion
Chapter 3 Pedestrian Detection Algorithm Based on Mobilenet-Tiny
3.1 Research Motivation
3.2 Improved YOLO v3 Tiny Based on Depthwise Separable convolutional Filter
3.3 Mobilenet-Tiny Model Training
3.3.1 Build Data Set
3.3.2 Mobilenet-Tiny Model Training
3.4 Experiment and Analysis
3.5 Conclusion
Chapter 4 Pedestrian Tracking Algorithm Based on Mobilenet-SORT
4.1 Research Motivation
4.2 Improved MOT Algorithm Mobilenet-SORT Based on SORT and Mobilenet-Tiny
4.2.1 SORT
4.2.2 Mobilenet-SORT
4.3 Experimental Process and Result Analysis
4.3.1 Data Processing
4.3.2 Experiment Design
4.3.3 Experimental Analysis
4.4 Conclusion
Chapter 5 Pedestrian Counting Based on Mobilenet-SORT and Implementation on Rasp-berry Pi
5.1 Research Motivation
5.2 Pedestrian Counting Based on Mobilenet-SORT
5.2.1 SORT Algorithm Counting Error Analysis Experiment
5.2.2 Mobilenet-SORT Pedestrian Counting Based on Regional Information
5.3 Implementation of Pedestrian Counting on Raspberry Pi
5.4 Conclusion
Chapter 6 Summary and Outlook
6.1 Summary
6.2 Outlook
References
致谢
Appendix A 中文摘要
本文编号：3341189
【文章来源】：华中师范大学湖北省 211工程院校 教育部直属院校
【文章页数】：80 页
【学位级别】：硕士
【文章目录】：
Abstract
Chapter 1 Introduction
1.1 Research Background and Significance
1.2 Research Status
1.3 Research Difficulties
1.4 Main Content and Thesis Structure
1.5 Conclusion
Chapter 2 Basic Principle
2.1 Overview
2.2 Principles Involved in Pedestrian Detection
2.2.1 YOLO v3
2.2.2 YOLO v3 Tiny
2.3 Principles Involved in Pedestrian Tracking
2.3.1 Intersection Over Union
2.3.2 Hungarian Algorithm
2.3.3 Kalman Filter
2.4 Related Concepts of Pedestrian Tracking Based on MOT
2.5 Conclusion
Chapter 3 Pedestrian Detection Algorithm Based on Mobilenet-Tiny
3.1 Research Motivation
3.2 Improved YOLO v3 Tiny Based on Depthwise Separable convolutional Filter
3.3 Mobilenet-Tiny Model Training
3.3.1 Build Data Set
3.3.2 Mobilenet-Tiny Model Training
3.4 Experiment and Analysis
3.5 Conclusion
Chapter 4 Pedestrian Tracking Algorithm Based on Mobilenet-SORT
4.1 Research Motivation
4.2 Improved MOT Algorithm Mobilenet-SORT Based on SORT and Mobilenet-Tiny
4.2.1 SORT
4.2.2 Mobilenet-SORT
4.3 Experimental Process and Result Analysis
4.3.1 Data Processing
4.3.2 Experiment Design
4.3.3 Experimental Analysis
4.4 Conclusion
Chapter 5 Pedestrian Counting Based on Mobilenet-SORT and Implementation on Rasp-berry Pi
5.1 Research Motivation
5.2 Pedestrian Counting Based on Mobilenet-SORT
5.2.1 SORT Algorithm Counting Error Analysis Experiment
5.2.2 Mobilenet-SORT Pedestrian Counting Based on Regional Information
5.3 Implementation of Pedestrian Counting on Raspberry Pi
5.4 Conclusion
Chapter 6 Summary and Outlook
6.1 Summary
6.2 Outlook
References
致谢
Appendix A 中文摘要
本文编号：3341189
本文链接：https://www.wllwen.com/kejilunwen/shengwushengchang/3341189.html
最近更新
教材专著
