-
Notifications
You must be signed in to change notification settings - Fork 6
/
bounding_boxes_worst.m
236 lines (196 loc) · 10.6 KB
/
bounding_boxes_worst.m
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
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
clear all; close all;
addpath Ensemble_Regressors/
addpath HelperFunctions/
load box_regression_coco4_alldata
IMG_ROOT = './Datasets/RealWorld/Moshik/coco_reg4_images/';
%% RUN ENSEMBLE METHODS FOR EVERY CLASS INDIVIDUALLY
classes = unique(class_id); % classes = 4; %
response_vars = {'x1','y1','x2','y2'}; % response_vars = {'x1'}; %
fprintf(['\tClass\t # Instances (n)\n']); disp([unique(class_id),histc(class_id,unique(class_id))]);
results_summary = {};
tests_size = [length(response_vars),length(classes)];
for cls_idx=1:length(classes)
clear PRED_UPCR PRED_ORIG TRUE_RESP MEAN_TRUE MEAN_PRED MEAN_X MEAN_Y X_* Y_* H_* W_*
for var_idx=1:length(response_vars)
% fprintf(['RESPONSE ' response_vars{var_idx} ', CLASS ' num2str(classes(cls_idx)) '\n==============================\n']);
clear Z y_*
Z = eval(['Z_' response_vars{var_idx}]); Z = Z(:,class_id == classes(cls_idx));
y_true = eval(['yy_true_' response_vars{var_idx}]); y_true = y_true(class_id == classes(cls_idx));
file_idx = sub2ind(tests_size,var_idx,cls_idx); % the index of the current experiment
%% START HERE
Z = Z([4,6,8,9,11,12],:); % high correlation between the 8 best regressors, this filters all of them [val loc]=sort(mean(MSE_orig,2)); imagesc(corr(Z(loc,loc)'));
%Z([9,11],:) = [];
orig_mean = mean(y_true); orig_mean_pred = mean(Z,2);
y_true = double(y_true) - mean(y_true);
Z = bsxfun(@minus, Z, mean(Z,2));
[m,n] = size(Z);
Ey = mean(y_true);
Ey2 = mean(y_true.^2);
var_y = Ey2 - Ey.^2;
C = cov(Z');
[v_1 lambda_1] = eigs(C,1,'lm');
rho_true = mean(Z .* repmat(y_true,m,1),2);
mse = @(x) mean((y_true' - x).^2 / var_y);
mse_true = zeros(m,1);
for i=1:m
mse_true(i) = mse(Z(i,:)');
end
if ~exist('MSE_orig','var')
MSE_orig = zeros(m,prod(tests_size)); %to access use: sub2ind(tests_size,var_idx,cls_idx)
G2_EST = zeros(prod(tests_size),1);
VAR_Y = zeros(prod(tests_size),1);
PRED_UPCR = zeros(n,length(response_vars));
PRED_ORIG = zeros(n,length(response_vars),m);
TRUE_RESP = zeros(n,length(response_vars));
MEAN_TRUE = zeros(1,length(response_vars));
MEAN_PRED = zeros(m,length(response_vars));
end
VAR_Y(file_idx) = var_y;
for i=1:m
MSE_orig(i, file_idx) = mse(Z(i,:)');
end;
MEAN_TRUE(1,var_idx) = orig_mean;
MEAN_PRED(:,var_idx) = orig_mean_pred;
[y_oracle2, w_oracle2] = ER_Oracle_2_Unbiased(y_true, Z);
[y_best,w_best] = ER_BestRegressor(y_true,Z);
[y_MEAN,beta_MEAN] = ER_MeanWithBiasCorrection(Z, Ey);
%y_biasedmean = mean(Z)'; % most teams are unbiased, so this should be equivalent to the mean
y_MED = ER_MedianWithBiasCorrection(Z, Ey);
% [y_DGEM,w_DGEM] = ER_UnsupervisedDiagonalGEM(Z, Ey);
% %y_gem = ER_UnsupervisedGEM(Z, Ey,Ey2);
% y_UPCR_delta0 = ER_SpectralApproach(Z, Ey, Ey2);
% [y_UPCR,w_UPCR] = ER_SpectralApproachGivenDeltaStar(Z, Ey, Ey2,mse(y_oracle2));
% [y_UPCR_MRE,w_UPCR_MRE] = ER_SpectralApproachDeltaMinMRE(Z, Ey, Ey2,mse(y_oracle2));
% [y_UPCR_WMRE,w_UPCR_WMRE] = ER_SpectralApproachDeltaMinWMRE(Z, Ey, Ey2,mse(y_oracle2));
% [y_IND,w_IND] = ER_IndependentMisfits(Z,Ey, Ey2);
% [y_UPCRt1,w_UPCRt1] = ER_SpectralApproachWeightsSum1(Z, Ey, Ey2);
%% Bayes Optimal Methods
[~, ~,~,~, G2_EST(file_idx)] = ER_IndependentMisfitsBayes(y_true, Z, Ey, Ey2,'l2',0, true); % use initialization to get G2_EST
is_bad_dataset = 1+double(G2_EST(file_idx)/var_y < .2);
% fprintf(is_bad_dataset,'G2_EST = %.2f\n',G2_EST(file_idx)/var_y);
[y_INDB, w_INDB,rho_INDB, MSE_hat_INDB] = ER_IndependentMisfitsBayes(y_true, Z, Ey, Ey2,'l2',0);
[y_UPCRrhoINDB, w_UPCRrhoINDB] = ER_UPCRgivenRho(Z,Ey,Ey2,rho_INDB);
[y_UPCRrhoOracle, w_UPCRrhoOracle] = ER_UPCRgivenRho(Z,Ey,Ey2,rho_true);
[y_UPCRrhoINDB2c, w_UPCRrhoINDB2c] = ER_UPCRgivenRho2Components(Z,Ey,Ey2,rho_INDB);
[inlier_idx,outlier_idx, MSE_ss] = subset_selection(y_true,Z,Ey,Ey2,'rho');
if sum(inlier_idx) > 4
[y_INDB_ss, w_INDB_ss,rho_INDB_ss, ~] = ER_IndependentMisfitsBayes(y_true, Z(inlier_idx,:), Ey, Ey2,'l2',0);
[y_UPCRrhoINDB_ss, w_UPCRrhoINDB_ss] = ER_UPCRgivenRho(Z(inlier_idx,:),Ey,Ey2,rho_INDB_ss);
[y_UPCRrhoINDB2c_ss, w_UPCRrhoINDB2c_ss] = ER_UPCRgivenRho2Components(Z(inlier_idx,:),Ey,Ey2,rho_INDB_ss);
else
% If after subset selection we have 4 or less inliers, take their average (that's not enough
% to rubstly estimate rho using least squares)
[y_INDB_ss, w_INDB_ss] = ER_MeanWithBiasCorrection(Z(inlier_idx,:), Ey);
[y_UPCRrhoINDB_ss, w_UPCRrhoINDB_ss] = ER_MeanWithBiasCorrection(Z(inlier_idx,:), Ey);
[y_UPCRrhoINDB2c_ss, w_UPCRrhoINDB2c_ss] = ER_MeanWithBiasCorrection(Z(inlier_idx,:), Ey);
end;
% %% MSE Results
% fprintf('\n\n');
% mse = @(x) mean((y_true' - x).^2) / var_y;
% for alg=who('y_*')'
% if ((~strcmp(alg{1}, 'y_true')) && (~strcmp(alg{1}, 'y_true_orig')))
% alg_name = alg{1}; alg_name = alg_name(3:end);%upper(alg_name(3:end));
% fprintf('%s\t%.3f\n',alg_name,(mse(eval(alg{1}))));
% end;
% end;
% fprintf('Best individual MSE: %g\n', (min(mean((Z - repmat(y_true,m,1)).^2,2))/var_y));
%% Rank Correlation
MSE_INDB = zeros(m,1);
for i=1:m
MSE_INDB(i) = (var_y - 2*rho_INDB(i) + C(i,i)) / var_y;
end;
mse_true_ss = mse_true(inlier_idx);
[mse_val, mse_loc] = min(MSE_INDB(inlier_idx));
EXCESS_MSE(file_idx) = mse_true_ss(mse_loc)-min(mse_true);
[rho_val, rho_loc] = max(rho_INDB(inlier_idx));
EXCESS_rho(file_idx) = mse_true_ss(rho_loc)-min(mse_true);
tmp = Z(inlier_idx,:);
y_BEST_MSEhat = tmp(mse_loc,:)';
y_BEST_RHOhat = tmp(rho_loc,:)';
PRED_UPCR(:,var_idx) = y_UPCRrhoINDB2c_ss + orig_mean;
for i=1:m
PRED_ORIG(:,var_idx,i) = Z(i,:)' + orig_mean_pred(i);
end;
TRUE_RESP(:,var_idx) = y_true' + orig_mean;
end; % FOR EACH RESPONSE VARIABLE (X1,Y1,X2,Y2)
%% CALCULATE SmoothL1Loss ON CENTER, WIDTH AND HEIGHT
X_TRUE = (TRUE_RESP(:,1)+TRUE_RESP(:,3))/2;
Y_TRUE = (TRUE_RESP(:,2)+TRUE_RESP(:,4))/2;
W_TRUE = TRUE_RESP(:,3)-TRUE_RESP(:,1);
H_TRUE = TRUE_RESP(:,4)-TRUE_RESP(:,2);
LOC_TRUE = [TRUE_RESP(:,1), TRUE_RESP(:,2), W_TRUE, H_TRUE];
%LOC_TRUE = [X_TRUE, Y_TRUE, W_TRUE, H_TRUE];
X_UPCR = (PRED_UPCR(:,1)+PRED_UPCR(:,3))/2;
Y_UPCR = (PRED_UPCR(:,2)+PRED_UPCR(:,4))/2;
W_UPCR = PRED_UPCR(:,3)-PRED_UPCR(:,1);
H_UPCR = PRED_UPCR(:,4)-PRED_UPCR(:,2);
LOC_UPCR = [PRED_UPCR(:,1), PRED_UPCR(:,2), W_UPCR, H_UPCR];
%LOC_UPCR = [X_UPCR, Y_UPCR, W_UPCR, H_UPCR];
X_ORIG = zeros(n,m); Y_ORIG = zeros(n,m); W_ORIG = zeros(n,m); H_ORIG = zeros(n,m);
LOC_ORIG = zeros(n,4,m);
for i=1:m
X_ORIG(:,i) = (PRED_ORIG(:,1,i)+PRED_ORIG(:,3,i))/2;
Y_ORIG(:,i) = (PRED_ORIG(:,2,i)+PRED_ORIG(:,4,i))/2;
W_ORIG(:,i) = PRED_ORIG(:,3,i)-PRED_ORIG(:,1,i);
H_ORIG(:,i) = PRED_ORIG(:,4,i)-PRED_ORIG(:,2,i);
LOC_ORIG(:,:,i) = [PRED_ORIG(:,1,i), PRED_ORIG(:,2,i), W_ORIG(:,i), H_ORIG(:,i)];
end;
%%
%SmoothL1Loss(x) =
% // 0.5 * (x) ** 2 -- if x < 1.0
% // |x| - 0.5-- otherwise
SmoothL1Loss = @(x) ((.5 * (x).^2) .* (x<1)) + ((abs(x) - .5).*(x>=1));
TotalLoss = @(x,y,w,h) (SmoothL1Loss(x-X_TRUE).*SmoothL1Loss(y-Y_TRUE).*SmoothL1Loss(w-W_TRUE).*SmoothL1Loss(h-H_TRUE));
loss_upcr = TotalLoss(X_UPCR,Y_UPCR,W_UPCR,H_UPCR);
loss_orig = zeros(n,m);
for i=1:m
loss_orig(:,i) = TotalLoss(X_ORIG(:,i),Y_ORIG(:,i),W_ORIG(:,i),H_ORIG(:,i));
end;
loss_mean = mean(loss_orig,2);
loss_prod = nthroot(prod(loss_orig,2),m);
%% look for high loss_mean/prod and low loss_upcr
[val1 loc1] = sort(loss_mean - loss_upcr,'descend');
[val2 loc2] = sort(loss_prod - loss_upcr,'descend');
% fprintf(['CLASS ' num2str(classes(cls_idx)) ':\n']);
d=squeeze(ALLDATA(1,:,7:10));
CLASS_DICT = {'person','chair','table','dog','cat','sofa'}; %(categories_id_dict={'person': 0, 'chair': 1, 'table': 2,'dog': 3,'cat': 4,'sofa': 5}
fprintf(['CLASS: ' CLASS_DICT{classes(cls_idx)+1} '\n']);
for i=1:5
%%
correction = [1 1 0 0];
image_id = ALLDATA(1,all([d(:,1) d(:,2) d(:,3) d(:,4)] == repmat(TRUE_RESP(loc1(i),:),size(d,1),1),2),1);
fprintf(['FIG1 IMAGE ID: ' num2str(image_id) ' IMAGE LOC: ' num2str(loc1(i)) '\n']);
im = imread([IMG_ROOT num2str(image_id) '.jpg']);
fig1=figure(1); clf; imshow(im); hold on;
for j=1:m
if all(LOC_ORIG(loc1(i),:,j) >=0)
rectangle('Position', LOC_ORIG(loc1(i),:,j) + correction,'EdgeColor','r', 'LineWidth', 2);
fprintf(['LOC_ORIG ' num2str(j) ': ']); LOC_ORIG(loc1(i),:,j),
end;
end;
rectangle('Position', LOC_UPCR(loc1(i),:) + correction,'EdgeColor','g', 'LineWidth', 3,'LineStyle','--');
fprintf('LOC_UPCR: '); LOC_UPCR(loc1(i),:),
rectangle('Position', LOC_TRUE(loc1(i),:) + correction,'EdgeColor','b', 'LineWidth', 3);
fprintf('LOC_TRUE: '); LOC_TRUE(loc1(i),:),
%rectangle('Position', mean(LOC_ORIG(loc1(i),:,:),3),'EdgeColor','g', 'LineWidth', 2);
print('-depsc', ['plots/icml/bbox_images/' CLASS_DICT{classes(cls_idx)+1} num2str(image_id) '.eps']);
image_id = ALLDATA(1,all([d(:,1) d(:,2) d(:,3) d(:,4)] == repmat(TRUE_RESP(loc2(i),:),size(d,1),1),2),1);
fprintf(['FIG2 IMAGE ID: ' num2str(image_id) ' IMAGE LOC: ' num2str(loc1(i)) '\n']);
im = imread([IMG_ROOT num2str(image_id) '.jpg']);
fig2=figure(2); clf; imshow(im); hold on;
for j=1:m
if all(LOC_ORIG(loc2(i),:,j) >=0)
rectangle('Position', LOC_ORIG(loc2(i),:,j) + correction,'EdgeColor','r', 'LineWidth', 2);
fprintf(['LOC_ORIG ' num2str(j) ': ']); LOC_ORIG(loc2(i),:,j),
end;
end
rectangle('Position', LOC_UPCR(loc2(i),:) + correction,'EdgeColor','g', 'LineWidth', 3,'LineStyle','--');
fprintf('LOC_UPCR: '); LOC_UPCR(loc2(i),:),
rectangle('Position', LOC_TRUE(loc2(i),:) + correction,'EdgeColor','b', 'LineWidth', 3);
fprintf('LOC_TRUE: '); LOC_TRUE(loc2(i),:),
%rectangle('Position', mean(LOC_ORIG(loc2(i),:,:),3),'EdgeColor','g', 'LineWidth', 2);
print('-depsc', ['plots/icml/bbox_images/' CLASS_DICT{classes(cls_idx)+1} num2str(image_id) '.eps']);
%%
%fprintf('PAUSE\n'); pause;
end
end; % FOR EACH CLASS