HAT_dataset compute accuracy for caffe_deep learning approach of matlab interface--->>> code record

<span style="font-family:SimHei;font-size:24px;">%%
clc;  close all;  clear all;
use_gpu=1; % GPU 模式

if exist('../+caffe', 'dir')
addpath('..');
else
error('Please run this demo from caffe/matlab/demo');
end

if exist('use_gpu', 'var') && use_gpu
caffe.set_mode_gpu();
gpu_id = 0;  % we will use the first gpu in this demo
caffe.set_device(gpu_id);
else
caffe.set_mode_cpu();
end

model_dir = '../../models/bvlc_alexnet/';
net_model = [model_dir 'train_val.prototxt'];
% net_model = [model_dir '__deploy.prototxt'];
net_weights = [model_dir 'caffe_alexnet_train_iter_310000.caffemodel'];
% net_weights = [model_dir 'bvlc_alexnet.caffemodel'];

phase = 'train';

if ~exist(net_weights, 'file')
error('Please download CaffeNet from Model Zoo before you run this demo');
end

% Initialize a network
net = caffe.Net(net_model, net_weights, phase);

fprintf('Extract Predict Scores ...\n');

% dataPath = '../../Link to test_code/traindataset/';
dataPath = '../../fine_tuning_data/HAT_fineTuning_data/test_data/';

images = dir([dataPath,'*.png']);

for num=1:length(images)

%     disp('read the', num2str(num) ,'th', '/', num2str(num1) ,'image, please waiting ...');
im = imread([dataPath, images(num).name]);
%     fid1 = fopen('train_ids.txt', 'a');
%     fprintf(fid1, '%s \n', images(num).name );
%     fclose(fid1);

train_image = im;
train_image = imresize(train_image, [320, 240]);
%     imshow(train_image);
image_name = images(num).name;
id = strtok(image_name,'_');
id = str2double(id);
im = train_image;

% prepare oversampled input
% input_data is Height x Width x Channel x Num
%     tic;
input_data = {prepare_image(im)};
%     toc;

% do forward pass to get scores  前向传播
% scores are now Channels x Num, where Channels == 1000
%     tic;
% The net forward function. It takes in a cell array of N-D arrays
% (where N == 4 here) containing data of input blob(s) and outputs a cell
% array containing data from output blob(s)

savePath = '/home/d302/wangxiao/caffe-master/wangxiao/bvlc_alexnet/';

scores = net.forward(input_data);
Score = scores{1,1};

tmp = zeros([27, 1]);
matrix = zeros([1900, 27]);

for i = 1:size(Score, 1)
all_vector = Score(i, :);
mean_vector = mean(all_vector);
tmp(i, 1) = mean_vector;
end
tmp = tmp';

fid = fopen([savePath, 'predict_score.txt'], 'a');
fprintf(fid, '%f ', tmp);
fprintf(fid, '\n');
fclose(fid);

%     toc;

%     scores = scores{1};
%     scores = mean(scores, 2);  % take average scores over 10 crops
%     [~, maxlabel] = max(scores);
%     call caffe.reset_all() to reset caffe
%     caffe.reset_all();

end

disp('extract predict score, done !');

predict_score = importdata([savePath, 'predict_score.txt']);
gt_data = '/home/d302/wangxiao/caffe-master/fine_tuning_data/HAT_fineTuning_data/test_data_fineTuning.txt';
gt_score = importdata(gt_data);

tmp_predict_score = predict_score ;

% for threshold = 0.4:0.05:0.55
for threshold = 0.5
for i = 1:size(tmp_predict_score, 1)
for j = 1:size(tmp_predict_score, 2)

if (tmp_predict_score(i, j) >threshold)
tmp_predict_score(i, j) = 1;
else
tmp_predict_score(i, j) = 0;
end
end
end
end

for j = 1:size(gt_score.data, 2)  % column

gt_vector = gt_score.data(:, j);
predict_vector = tmp_predict_score(:, j);

TP = 0;   FP = 0;   TN = 0;   FN = 0;

for i = 1:size(gt_score.data, 1)
true_label = gt_vector(i);
predict_label = predict_vector(i);

if (predict_label==1) && (true_label==1)
TP =TP+1;
end
if (predict_label==1) && (true_label==0)
FP = FP+1;
end
if (predict_label==0) && (true_label==0)
TN = TN+1;
end
if (predict_label==0) && (true_label==1)
FN = FN+1;
end

end

all_positive = TP + FP;    all_negative = TN + FN;
accuracy = 0.5*(TP/all_positive + TN/all_negative);
disp([num2str(accuracy)]);

end</span>