Design And Implementation Of Android Application For Local Album Classification Based On Tensorflow

The development of mobile applications has developed rapidly in recent years,and almost all the functions commonly used by users have already had corresponding software to meet the demand.At the same time,however,the user's requirements for the application are correspondingly increased,not just for the function implementation.In addition to breakthroughs in functionality and performance,mobile applications have increased user privacy and security requirements.How to develop high-performance and secure applications in a more efficient and reasonable way to meet the needs of users has become the top priority for every mobile developer.At the same time,artificial intelligence based on deep learning has developed rapidly in recent years,which has made a major breakthrough in many technical fields.How to mobile development have a new way of trying by effectively combining with artificial intelligence.The traditional image classification application needs to upload the image to the cloud server through the network,and then perform image classification and re-send the classification label to the mobile terminal,and then the mobile terminal displays the classification result according to the label.Such an approach is not only dependent on network,it is inefficient and has security risks.In order to overcome the above problems,this paper designs and implements an android application software that can classify user albums locally on the mobile terminal.The software classifies the user's photos by running an image classification neural network on the mobile terminal through the Tensorflow neural network framework.The main work and results of this paper are as follows:(1)For the development of Android applications,introduce the basic architecture of Android,image caching technology,database technology and Android system reference to third-party libraries.(2)For the implementation of image classification function,the Tensorflow training framework and CNN image classification network technology are introduced.(3)For the subsequent development and functional expansion of the application,the entire application is based on a modular design and related implementation.(4)The whole image classification system was designed,including image classification neural network training system and image classification software system,and related functions were designed and implemented.(5)The overall performance test and functional verification of the image classification system are carried out.(6)Summarize the design and implementation of the image classification system in the text,and propose the direction of subsequent function expansion and improvement.