train_id_net_res_2stream_gan.m

function train_id_net_res_2stream(varargin)
% -------------------------------------------------------------------------
% Part 4.1: prepare the data
% -------------------------------------------------------------------------

% Load character dataset
imdb = load('./url_data_gan_24000.mat') ;
imdb = imdb.imdb;
imdb.images.set(:) = 1;
% -------------------------------------------------------------------------
% Part 4.2: initialize a CNN architecture
% -------------------------------------------------------------------------
net = resnet52_2stream_gan();
net.conserveMemory = true;
net.meta.normalization.averageImage = reshape([105.6920,99.1345,97.9152],1,1,3);

% -------------------------------------------------------------------------
% Part 4.3: train and evaluate the CNN
% -------------------------------------------------------------------------
opts.train.averageImage = net.meta.normalization.averageImage;
opts.train.batchSize = 32;
opts.train.continue = true; 
opts.train.gpus = 3;
opts.train.nesterovUpdate = true ;
opts.train.prefetch = false ;
opts.train.expDir = './data/resnet52_2stream_drop0.9_baseline_batch32_gan24000_all';  % your model will store here
opts.train.learningRate = [0.1*ones(1,40),0.01*ones(1,20)];
opts.train.derOutputs = {'objective', 0.5,'objective_2', 0.5,'objective_final', 1} ;
opts.train.weightDecay = 0.0001;
opts.train.numEpochs = numel(opts.train.learningRate) ;
[opts, ~] = vl_argparse(opts.train, varargin) ;

% Call training function in MatConvNet
[net,info] = cnn_train_dag(net, imdb, @getBatch,opts) ;


% --------------------------------------------------------------------
function inputs = getBatch(imdb, batch,opts)
% --------------------------------------------------------------------
im1_url = imdb.images.data(batch) ; 
label1 = imdb.images.label(:,batch) ;
batchsize = numel(batch);
% 1 for true,2 for different
batch2 = zeros(batchsize,1);
half = round(batchsize/2);
label_f = cat(1,ones(half,1,'single'),ones(batchsize-half,1,'single')*2);
for i=1:batchsize
    if(i<=half)
        batch2(i) = rand_same_class(imdb, batch(i));
    else
        batch2(i) = rand_diff_class(imdb, batch(i));
    end
end
im2_url = imdb.images.data(batch2) ; 
im1 = vl_imreadjpeg(im1_url,'Pack','Resize',[224,224],'Flip',...
    'CropLocation','random','CropSize',[0.85,1],'Interpolation', 'bicubic','NumThreads',8);
im2 = vl_imreadjpeg(im2_url,'Pack','Resize',[224,224],'Flip',...
    'CropLocation','random','CropSize',[0.85,1],'Interpolation', 'bicubic','NumThreads',8);
label2 = imdb.images.label(:,batch2) ;

oim1 = bsxfun(@minus,im1{1},opts.averageImage); 
oim2 = bsxfun(@minus,im2{1},opts.averageImage); 
label_f(label1==0) = 0;
inputs = {'data',gpuArray(oim1),'data_2',gpuArray(oim2),'label',label1,'label_2',label2,'label_f',label_f'};