摘要
对手机中的所有用户图像进行分类,可以让用户更好地检索自己想要的图片,大量的App应用通过服务器对相册进行分类,而利用性能相对羸弱的移动端设备本身,达到类似的分类效果是亟待解决的问题。TensorflowLite框架是能够在Android端良好运行的神经网络训练框架。通过构建一个Android应用,基于TensorflowLite框架,移植经典图像处理的卷积神经网络进行图像相册分类。实验表明,应用程序可以在较短时间内处理图片并完成相册分类。
Classifying all the user images in the phone allows the user to better retrieve the images they want.A large number of App applications classify albums through servers,and how to use the relatively weak mobile device itself to achieve similar classification effect is an urgent problem to be solved.The TensorflowLite framework is a neural network training framework that can work well on the Android side.By building an Android App based on the TensorflowLite framework,classic image processing convolutional neural network can be transplanted for image album classification experiments show that the application can process images and complete album classification in a short period time.
作者
彭宏
庄宁
PENG Hong;ZHUANG Ning(College of Information Engineering,Zhejiang University of Technology,Hangzhou 310023,China)
出处
《浙江工业大学学报》
CAS
北大核心
2020年第2期165-172,共8页
Journal of Zhejiang University of Technology
基金
国家自然科学基金资助项目(61871349)。
