摘要
通过手机拍摄显示屏窃取敏感信息是导致信息泄漏最普遍的方式。针对此行为,介绍自主研发的基于人工神经网络的物体识别系统。能够实现对手机拍摄行为的实时监测,通过控制智能板卡隐藏显示内容保障信息安全。经测试,将神经网络YOLOv4-Tiny原型特征提取网络增加一个有效特征层进行预测,能显著地提高对手机拍摄行为智能识别的精度。为此着重论述对手机拍摄行为识别的优化思路并验证生成模型识别的可靠性,作为以后在更多领域实践的参考。
Stealing sensitive information through the screen of a mobile phone is the most common way to cause information leakage.In response to this behavior,the independently-developed object recognition system based on ANN(artificial neural network)is introduced.The system can realize real-time monitoring of mobile phone photograph behavior,and ensure information security by controlling the smart board to hide and display content.The experiment and test indicate that by adding an effective feature layer to the ANN YOLOv4-Tiny prototype feature extraction network for prediction,the accuracy of intelligent recognition of mobile phone photograph behavior can be significantly improved.This paper focuses on the optimization ideas for mobile phone photograph behavior recognition and the reliability verification of generative model recognition,as a reference for future practice in many fields.
作者
王昕
寇云峰
辛浪
宋滔
WANG Xin;KOU Yunfeng;XIN Lang;SONG Tao(China Cyber Security,Chengdu Sichuan 610041,China)
出处
《通信技术》
2021年第1期244-250,共7页
Communications Technology
关键词
大数据
人工神经网络
深度学习
智能识别
信息安全
big data
ANN(artificial neural network)
deep learning
intelligent recognition
information security
