###
计算机系统应用:2018,27(8):28-34
←前一篇   |   后一篇→
本文二维码信息
码上扫一扫!
基于深度学习的运动目标实时识别与定位
童基均, 常晓龙, 赵英杰, 蒋路茸
(浙江理工大学 信息学院, 杭州 310018)
Real-Time Detection and Positioning of Moving Target Based on Deep Learning
TONG Ji-Jun, CHANG Xiao-Long, ZHAO Ying-Jie, JIANG Lu-Rong
(School of Information Science and Technology, Zhejiang Sci-Tech University, Hangzhou 310018, China)
摘要
图/表
参考文献
相似文献
本文已被:浏览 208次   下载 178
投稿时间:2017-12-01    修订日期:2017-12-21
中文摘要: 针对人体运动目标的实时检测与定位问题,采用深度学习的方法进行研究.在Caffe框架下,采用SSD (Single Shot multibox Detector)检测方法.以VGG16作为基础网络模型,增加额外特征卷积层,提取多尺度的卷积特征.然后对实验数据集进行迭代训练,得到运动目标检测模型.利用训练好的模型,通过2路摄像机检测运动目标,并双目视觉定位.实验结果表明,整个系统运行速度可达40 fps,在10 m×10 m的场景下,平均定位误差在6 cm以内,在速度和精度上均有很好的表现,为大中型场景的人体运动实时检测定位问题提供了有效的解决方案.
Abstract:Aiming at the issues of real-time detection and positioning of movement target, a method of deep learning is proposed. The Single Shot multibox Detector (SSD) detection method is used under Caffe framework, the VGG16 model is used as the basic network model, and the additional feature convolution layers are used to extract the multi-scale convolution features. Then the experimental data set is iteratively trained to get the motion target detection model. The moving objects are detected using the trained model and then positioned through binocular vision positioning method. The experiment results show that the system can reach 40 fps. In the 10 m×10 m scene, the average positioning error is within 6 cm. The system has sound performance both in speed and precision, which provides an effective solution for the real-time detection and positioning of human motion in large and medium-sized scenes.
文章编号:     中图分类号:    文献标志码:
基金项目:浙江省重点研发计划(2015C03023);浙江理工大学“521人才培养计划”
引用文本:
童基均,常晓龙,赵英杰,蒋路茸.基于深度学习的运动目标实时识别与定位.计算机系统应用,2018,27(8):28-34
TONG Ji-Jun,CHANG Xiao-Long,ZHAO Ying-Jie,JIANG Lu-Rong.Real-Time Detection and Positioning of Moving Target Based on Deep Learning.COMPUTER SYSTEMS APPLICATIONS,2018,27(8):28-34

用微信扫一扫

用微信扫一扫