superpixel-owaoc/adapImpTrain.m at master · carolinepacheco/superpixel-owaoc · GitHub

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
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
function [A, impFeatureTime] = adapImpTrain(varargin)

label_test = struct([]);
impF = struct([]);
aPool = struct([]);
impFeatureTime  = struct([]);
clabel = struct([]);
disResult  = struct([]);

for k = 1:size(varargin{1},1)
    X = (sprintf('Adaptation features to superpixel: %d',k));
    disp(X);

    %%%%%%COLOR DATA ADPAT
    TestDataGray = varargin{1}(k,:);
    TestDataRed = varargin{2}(k,:);
    TestDataGreen = varargin{3}(k,:);
    TestDataBlue = varargin{4}(k,:);

    %%%%%%TEXTURE DATA ADPAT
    TestDataXCS = varargin{5}(k,:);

    %%%%%%GT DATA ADPAT
    dataGT = varargin{6}(k,:);
    dataGT = dataGT';

    TestData = [TestDataGray' TestDataRed' TestDataGreen' TestDataBlue' TestDataXCS'];

   for i = 1:size(TestData,1) %verificando o sample individualmente

   for c= 1:size(TestData,2) %check number features/clqssificadores

      TestD = TestData(:, c);
      w0 = varargin{7}{k}{1,c}{1,1};
      outColor = TestD*w0;
      label_test{c} = outColor*labeld;
      disResult{c} = +outColor;
   end %final check number features

   label_target = 0;
   label_outlier = 0;
   label = 0;
   re_label = 0;

for c= 1:size(TestData,2) %check number features/clqssificadores 2

  data = TestData(i,c);
  dataDist = zeros(1, size(data,2));
  distance = disResult{c};

            if distance(i,1) > distance(i,2)
                dataDist(1,1:size(data,2)) = distance(i,1);
                %euclidian distance
                D = pdist2(data,dataDist,'euclidean');

                prob =  exp(-(D)/0.5); %melhor configuracao

                if prob >= 0.95

                    label_outlier = label_outlier + 1;
                    label = 'outlier';
                    re_label = -1;
                else
                    label_target = label_target + 1;
                    label = 'target ';
                    re_label = 1;
                end
            end

            if distance(i,1) < distance(i,2)
                dataDist(1,1:size(data,2)) = distance(i,2);
                D = pdist2(data,dataDist,'euclidean');
                prob =  exp(-(D)/0.5); %melhor configuracao

                if prob >= 0.95
                    label_target = label_target + 1;
                    label = 'target ';
                    re_label = 1;

                else
                    label_outlier = label_outlier + 1;
                    label = 'outlier';
                    re_label = -1;
                end
            end
            [accuracyClassPool] = calc_accuracy(label,dataGT(i));
            aPool{k}{c} = accuracyClassPool; %accuracy calculation each base classifier
            clabel{k}{c} = re_label;

      end

        if (i == 1)
           [impFeature] = adapImCalInitial(size(TestData,2),aPool{k});
        end


   THx = 0;
   strongH = 0;
   for c= 1:size(TestData,2) %check number features/classificadores

      Hx = impFeature{c}*clabel{k}{c};
      THx = THx + Hx;
   end %final check number features


      %FINAL CLASSIFICATION
      strongH = THx/c;
      if strongH >= 0
          labelC = 'target ';
          label_out(i) = 0;

      else
          labelC = 'outlier';
          label_out(i) = 1;

      end


      [accuracyClas] = calc_accuracy(labelC,dataGT(i));

      %adaptative feature importanc calculation
      [impFeature]= adapImpCal(size(TestData,2), aPool{k}, accuracyClas, impFeature);

      %guarda para cada amostra
      impF{k} = impFeature;
       impFeatureTime{i} = impF;

  end
  clearvars impFeature
  clearvars aPool
  clearvars sPool
  clearvars accuracyPool
  M(k,:) = label_out;
end

%[A] = regionToPixel2(M, varargin{14});%checar essas variavei
[A] = regionToPixel(M, varargin{8});

disp('END Adaptation Features');
end