本文選題：深度學(xué)習(xí) + 圖像分類　； 參考：《東華理工大學(xué)》2017年碩士論文

【摘要】：近幾年以來(lái),隨著科學(xué)計(jì)算機(jī)網(wǎng)絡(luò)及人工智能領(lǐng)域的發(fā)展,圖形圖像數(shù)據(jù)量逐漸增多,于是,如何從大量的自然圖像中快速提取到視覺(jué)特征已經(jīng)成了機(jī)器智能學(xué)習(xí)中的熱點(diǎn)研究課題,進(jìn)而對(duì)自然圖像的分類必然成為獲取自然圖像信息的研究重點(diǎn)。卷積神經(jīng)網(wǎng)絡(luò)是深度學(xué)習(xí)在圖像處理方面的一個(gè)重要應(yīng)用。它相比于其它機(jī)器學(xué)習(xí)算法如SVM等,其優(yōu)點(diǎn)是能夠直接對(duì)圖像像素進(jìn)行卷積并提取特征,也能夠利用海量的圖像數(shù)據(jù)將網(wǎng)絡(luò)參數(shù)訓(xùn)練充分,以達(dá)到更好的分類效果。本文對(duì)基于深度學(xué)習(xí)的自然圖像分類方法展開(kāi)研究,主要工作及創(chuàng)新點(diǎn)如下:1)基于tensorflow深度學(xué)習(xí)框架平臺(tái)設(shè)計(jì)一個(gè)用于識(shí)別圖像的淺層卷積神經(jīng)網(wǎng)絡(luò),并分別用單GPU和多GPU訓(xùn)練加速來(lái)對(duì)比該網(wǎng)絡(luò)性能,其中多GPU訓(xùn)練該網(wǎng)絡(luò)的所用的時(shí)間比單GPU縮短了25分鐘。該項(xiàng)工作的設(shè)計(jì)旨在建立一個(gè)較好的網(wǎng)絡(luò)結(jié)構(gòu)進(jìn)行訓(xùn)練和評(píng)估,并為工作3中建立更加復(fù)雜的網(wǎng)絡(luò)模型做鋪墊。2)本文圍繞卷積神經(jīng)網(wǎng)絡(luò)的網(wǎng)絡(luò)結(jié)構(gòu)和多參數(shù)分別進(jìn)行了改進(jìn)和優(yōu)化。研究實(shí)驗(yàn)表明,對(duì)batch值、dropout、momentum動(dòng)量值、數(shù)據(jù)集擴(kuò)增等的優(yōu)化,能夠有效地提高深層卷積神經(jīng)網(wǎng)絡(luò)模型的識(shí)別率。因此,合理的增加網(wǎng)絡(luò)層數(shù),優(yōu)化訓(xùn)練參數(shù)提高訓(xùn)練效率,以達(dá)到最佳的分類效果是圖像分類應(yīng)用研究中非常重要的目的。3)基于tensorflow深度學(xué)習(xí)框架平臺(tái),并用GPU訓(xùn)練加速來(lái)進(jìn)行卷積神經(jīng)網(wǎng)絡(luò)的網(wǎng)絡(luò)結(jié)構(gòu)改進(jìn)設(shè)計(jì)和參數(shù)優(yōu)化。首先,設(shè)計(jì)一個(gè)具有9層結(jié)構(gòu)的深層卷積神經(jīng)網(wǎng)絡(luò)。其次,用該網(wǎng)絡(luò)結(jié)構(gòu)分別對(duì)cifar-10和cifar-100等復(fù)雜圖像數(shù)據(jù)庫(kù)進(jìn)行訓(xùn)練、測(cè)試和優(yōu)化參數(shù)。結(jié)果表明,該網(wǎng)絡(luò)結(jié)構(gòu)相比之前研究者的網(wǎng)絡(luò)模型(Conv-KN)對(duì)這兩種復(fù)雜的圖像庫(kù)的分類準(zhǔn)確率分別提高了9.26%和3.55%。tensorflow框架平臺(tái)下的深層卷積神經(jīng)網(wǎng)絡(luò)的分類效果要明顯好于其它平臺(tái),并且tensorflow框架平臺(tái)下的訓(xùn)練時(shí)間上也得到了極大的提高。
[Abstract]:In recent years, with the development of scientific computer network and artificial intelligence, the amount of graphic and image data is increasing gradually. So how to quickly extract visual features from a large number of natural images has become a hot topic in machine intelligent learning, and then the classification of natural images will inevitably become the acquisition of natural image information. Research emphasis. Convolution neural network is an important application of deep learning in image processing. Compared with other machine learning algorithms, such as SVM, it has the advantage that it can convolution the image pixels directly and extract the features, and can also use massive image data to train the network parameters fully in order to achieve better classification results. This paper studies the classification method of natural image based on deep learning. The main work and innovation are as follows: 1) based on the tensorflow deep learning framework platform, a shallow convolution neural network for identifying images is designed, and the performance of the network is compared with single GPU and multiple GPU training respectively, in which the multi GPU is used to train the network. The time is 25 minutes shorter than the single GPU. The design of the work is designed to build a better network structure for training and evaluation, and to build a more complex network model for work 3. This paper improves and optimizes the network structure and multi parameters of the convolution neural network respectively. The research experiments show that the batch value, D The optimization of ropout, momentum momentum, and data set amplification can effectively improve the recognition rate of the deep convolution neural network model. Therefore, it is very important for the image classification application to increase the number of network layers, optimize the training parameters and improve the training efficiency, and to achieve the best classification effect in the study of the image classification application. Based on the depth of tensorflow, the.3 Study frame platform and accelerate the network structure improvement design and parameter optimization of convolution neural network with GPU training. First, a deep convolution neural network with 9 layers structure is designed. Secondly, the network structure is used to train, test and optimize the parameters of complex image databases such as cifar-10 and cifar-100, respectively. Compared with the previous researchers' network model (Conv-KN), the classification accuracy of the two complex image bases is improved by 9.26% and the deep convolution neural network under the 3.55%.tensorflow framework platform is better than the other platforms, and the training time under the tensorflow frame platform is also greatly improved. Improve.
【學(xué)位授予單位】：東華理工大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類號(hào)】：TP391.41;TP18

【參考文獻(xiàn)】

相關(guān)期刊論文 前2條

1 施智平;李清勇;趙曉東;何清;史忠植;;基于內(nèi)容圖像檢索中的優(yōu)化鑒別特征[J];計(jì)算機(jī)輔助設(shè)計(jì)與圖形學(xué)學(xué)報(bào);2012年12期

2 賈淑華;李星野;馬琳琳;姜興乾;;基于相關(guān)關(guān)系的圖像分類和圖像檢索[J];微計(jì)算機(jī)信息;2009年15期

相關(guān)博士學(xué)位論文 前1條

1 楊同峰;基于空間關(guān)系的圖像檢索與分類研究[D];山東大學(xué);2013年

，

本文編號(hào)：1915953