摘要
提出一种计算机视觉下半监督学习预警的人员踩踏风险,采用计算机视觉技术实现对数据的采集和处理,采用图像分割处理技术实现区域划分,在此基础上,通过半监督学习的方法,对实时采集的数据进行监督和不断更新的学习,从而达到实时预警的目的。最后采用一组非均衡分布下的人员密度进行系统预警的测试实验,结果显示,采用半监督学习方法,可以很好的实现计算机视觉系统下的人员踩踏风险预警,对高风险和低风险进行标定,实现预警,具有很好的安全使用价值。
The warning of staff stampede risk with computer vision technology under semi-supervised learning was proposed, the computer vision technology was used for data collection and processing, the image segmentation processing technology was used to do implement zoning, on this basis, through the semi-supervised learning method, the real-time datawas collected to monitor and update the study, and the purpose of real-time early warning was achieved. Finally, a nonequilibriumdistribution of the people was taken as target to test the ability, the result shows that with semi-supervisedlearning method, a good computer vision system personnel stampede risk warning is achieved, and high-risk and low-riskcalibrated was divided, it has good value for safety.
出处
《科技通报》
北大核心
2014年第4期128-130,共3页
Bulletin of Science and Technology
关键词
计算机视觉
半监督学习
人员踩踏预警
computer vision technology
semi-supervised learning
warning of staff stampede risk
