基于深度学习的视觉感知方法研究.pdf

本科毕业设计（论文）学院(部)机电工程学院题目基于深度学习的视觉感知方法研究电气工程及其年级2014级专业自动化班级14电气学号1429402058姓名朱永南指导老师陈良职称副教授论文提交日期2018年5月19日苏州大学本科生毕业设计（论文）目录摘要.........................................................1Abstract......................................................2第一章绪论..................................................31.1研究背景................................................31.2研究目的及意义..........................................31.3目标检测................................................31.4语义分割................................................51.5本文的主要工作..........................................6第二章深度学习及移动机器人视觉感知系统概述..................82.1深度学习...............................................82.2卷积神经网络...........................................82.3常用框架..............................................102.4移动机器人的视觉感知系统..............................122.5本章小结..............................................13第三章基于深度学习的目标检测模型...........................143.1R-CNN................................................143.2SPPNet...............................................143.3FastR-CNN............................................153.4FasterR-CNN..........................................153.5YOLO................................................16苏州大学本科生毕业设计（论文）3.6本章小结..............................................17第四章基于深度学习的语义分割模型...........................184.1ENet..................................................184.2ICNet.................................................194.3SemanticFusion.........................................204.4本章小结..............................................21第五章实验验证与分析.......................................225.1实验数据集............................................225.2运行环境..............................................235.3评判指标..............................................235.4目标检测模型实验......................................245.5语义分割模型实验......................................265.6在移动机器人上的实时目标检测..........................295.7本章总结..............................................32第六章总结与展望............