Development And Implementation Of Machine Learning Models Based On React Native

In recent years,artificial intelligence(AI)has become one of the hot-spots in the field of computer science.AI analyzes and comprehends the data that contains massive information.It enhances human's ability to improve the environment,govern the society and accomplish various tasks efficiently through intelligent algorithms,which is the ultimate pursuit of artificial intelligence.At present,with a sharp increase of image data and rapid advancement of computing power,Convolutional Neural Network(CNN),as a representative of Deep Learning(DL),has been applied to many computer vision-related-tasks,such as object detection,image classification,image segmentation and so on.Deep learning also archives a huge improvement compared to traditional algorithm,and has been successfully applied to many scenarios,including security,medical treatment,public transportation and education,which brings far-reaching social significance.In the field of deep learning,on the one hand,we must improve the accuracy and speed of algorithms,and keep making breakthroughs of the performance.On the other hand,we should also consider how to apply the deep learning technology to real-world scenarios thus to provide convenience for human life.Along with the rapid development of Internet industry,the number of mobile network clients is increasing year by year.Among many development frameworks,React Native has been widely sought since it is a mobile-client development framework truly realizing cross-platform.Based on the life-cycle principles of components,the React Native development framework renders the user interface according to the updates of attributes and states.It has performance as good as native applications,as well as cross-platform and real-time-loading capabilities,which can save a fair amount of cost and improve efficiency during development.This paper brings the algorithm of estimating haze index based on image,and two kinds of implementation of deep learning technology on the mobile-client,based on React Native development framework.This paper mainly contains three parts:1.Built on the principle of haze removal,this paper brings an algorithm of estimating PM2.5 based on image,which combines the image classification results and transmission calculations.2.In accordance with estimating PM2.5 of image,this paper completes implementation of deep learning model which has large-volume construction,high precision and low-speed requirement.The implementation on the mobile client is set up by cloud communication mode,including two mobile development systems of iOS and Android based on the React Native development framework.3.In accordance with fast object positioning and classification tasks,this paper completes implementation of deep learning model which has small-volume construction,regular precision and high-speed requirement.The implementation on the mobile client is set up by local communication mode,including two mobile development systems of iOS and Android based on the React Native development framework.