摘要
近年来，随着卷积神经网络理论的快速发展和移动设备的普及，基于移动式TensorFlow平台构建定制化的卷积神经网络分类模型，成为深度学习爱好者的主要研究方向之一。将个人图片库作为训练集输入模型，通过调整阀值和权值，最终获得定制化的分类模型，满足个性化的分类需求。对移动式TensorFlow平台、卷积神经网络及其移植到移动平台上的相关步骤进行了研究。通过此开发流程，为进一步使用移动式TensorFlow平台解决现实图片分类问题提供了参考。
In recent years, with the popularity of the rapid development of the theory of convolutional neural network and mo- bile devices, convolutional neural network classification model to build customized mobile platform Based on TensorFlow, has become one of the main research directions of deep learning enthusiasts. The personal image library is used as the input model of training set, and the customized classification model is finally obtained by adjusting the threshold and weight, which meets the personalized classification requirements. The related steps of mobile TensorFlow platform, convolutional neural network and its porting to mobile platform are studied. Through the development process, it provides a reference for the further use of mobile TensorFlow platform to solve the problem of realistic picture classification.
作者
李河伟
LI He-wei （ Beijing Wuzi University, Beijing 101149, China）
出处
《电脑知识与技术》
2017年第8期179-182,共4页
Computer Knowledge and Technology
