基于深度学习的起重机吊运手势识别  

Crane's lifting gestures recognition based on deep learning

作　　者：汪涛 张建德[2] WANG Tao;ZHANG Jiande(Center of Information and Network,Nanjing Medical University,Nanjing 210029,China;School of Computer Engineering,Nanjing Institute of Technology,Nanjing 211167,China)

机构地区：[1]南京医科大学信息与网络中心,江苏南京210029 [2]南京工程学院计算机工程学院,江苏南京211167

出　　处：《现代电子技术》2018年第23期127-129,133,共4页Modern Electronics Technique

基　　金：江苏省自然科学基金青年基金项目(BK20150731);江苏省高校自然科学基金面上项目(15KJB520014)~~

摘　　要：起重机工作环境嘈杂,指挥员的吊运手势在作业现场发挥着重要的作用,设计一个自动化吊运手势识别系统非常必要。以深度学习算法中卷积神经网络为基础方法,构建由静态手势模块和动态手势模块组成的复合神经网络模型,提取吊运手势的特征,最后对特征进行组合并识别。实验结果表明,该系统能够有效地识别吊运手势。The crane′s working environment is very noisy,and the gesture of commander plays an important role in the job site.Therefore,it is necessary to design an automatic lifting gesture recognition system.The convolution neural network in deep learning algorithm is used as the basic method.The combined neural network model composed of dynamic gesture module and static gesture module is constructed.The features of lifting gestures are extracted,combined and recognized.The experimental results show that the system can recognize the lifting gesture effectively.

关 键 词：吊运手势 卷积神经网络 手势分割 深度学习 颜色空间 特征提取 

分 类 号：TN99-34[电子电信—信号与信息处理] TP391.41[电子电信—信息与通信工程]