Research And Application Of Multi-layer Convolution Neural Network

In recent years,Deep learning technology in the field of computer vision,Natural Language Processing and speech recognition has surpassed the existing Traditional shallow machine learning technology in dealing with various problems.The emergence of Deep learning has revolutionized the development of Machine Learning and Computer Vision.Multi-layer convolution neural network(CNN)is considered as one of the most classical and powerful algorithms in the field of computer vision.Especially in image classification and target detection,multi-layer convolutional neural network has been accepted by academia and industry,and has been deployed in Google,Facebook,Amazon and Baidu used to solve the automatic image recognition and image annotation problems.In the learning process of image data the multi-layer convolution neural network involves a large number of optimization algorithms and training skills in machine learning,The main work of this paper is as follows:(1)Through the study of different variants of multi-layer convolutional neural networks,the traditional Le Net-5 model is replaced by the activation function and output function in the traditional Le Net-5 model,and the image classification is obtained on the lightweight data set of MNIST Good results.(2)For the classic Alex Net model,this paper introduces the batch standardization processing operation,and the image classification effect on the lightweight data set of CFIAR-10 is improved.(3)A colon polyp detection algorithm based on spatial continuity was proposed for colon polyp detection.Darknet-53 was used as the basic classification network.Based on the traditional deep convolutional neural network,the residual network,the upsampling layer,and the concat layer,multi-scale classification regression,and multiple classifiers improve detection accuracy.