matconvnet 테스트 이미지에 딥러닝모델 적용하고, 피쳐 뽑아 SVM으로 분류하는 예제

http://www.ninadthakoor.com/2016/07/25/getting-started-with-matconvnet/

 

Available at: http://www.vlfeat.org/matconvnet/

My goal is to use this toolbox to classify cars into four classes: Sedan, Minivan, SUV and pickup. I already have the data from my prior work (http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=6558793).

My plan is to first use transfer learning i. e. to use one of the deep networks pre-trained with imagenet data to extract features and use an SVM classifier to do the actual classification. I plan to do this with Matlab 2016a on Windows 10 PC equipped with GeForce GTX 960. I am aware that Matlab also has deep learning support in its Neural Network toolbox, I am going with MatConNet in hopes that it will stay more cutting edge.

The code below is derived from http://www.mathworks.com/company/newsletters/articles/deep-learning-for-computer-vision-with-matlab.html and http://www.vlfeat.org/matconvnet/quick/

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64

clear; close all; clc; %Load the pre-trained net net = load('imagenet-vgg-f.mat'); net = vl_simplenn_tidy(net) ;   %Remove the last layer (softmax layer) net.layers = net.layers(1 : end - 1);   %% This deals with reading the data and getting the ground truth class labels   %All files are are inside the root root = 'C:\Users\ninad\Dropbox\Ninad\Lasagne\data\c200x200\'; Files = dir(fullfile(root, '*.bmp'));   %Load the map which stores the class information load MakeModels.mat for i = 1 : length(Files)     waitbar (i/ length(Files));     % Read class from the map     Q = Files(i).name(end - 18 : end - 4);     Qout = MakeModels(Q);     Files(i).class = Qout.Type;           % Preprocess the data and get it ready for the CNN     im = imread(fullfile(root, Files(i).name));     im_ = single(im); % note: 0-255 range     im_ = imresize(im_, net.meta.normalization.imageSize(1:2));     im_ = bsxfun(@minus, im_, net.meta.normalization.averageImage);       % run the CNN to compute the features     feats = vl_simplenn(net, im_) ;     Files(i).feats = squeeze(feats(end).x); end   %% Classifier training   %Select training data fraction   trainFraction = 0.5; randomsort = randperm(length(Files)); trainSamples = randomsort(1 : trainFraction * length(Files)); testSamples = randomsort(trainFraction * length(Files)+1 : end);   Labels = [Files.class]; Features = [Files.feats];   trainingFeatures = Features( :, trainSamples); trainingLabels = Labels( :, trainSamples);   classifier = fitcecoc(trainingFeatures', trainingLabels);   %% Carry out the validation with rest of the data testFeatures = Features( :, testSamples); testLabels = Labels( :, testSamples);   predictedLabels = predict(classifier, testFeatures');   confMat = confusionmat(testLabels, predictedLabels);   % Convert confusion matrix into percentage form confMat = bsxfun(@rdivide,confMat,sum(confMat,2))   % Display the mean accuracy mean(diag(confMat))