forked from happynear/FaceDatasets
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathalign_megaface_from_list.m
162 lines (147 loc) · 6.12 KB
/
align_megaface_from_list.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
%% Align megaface with the help of the provided 3 points.
%% I got 739,181 out of 1M aligned faces using this code.
%% The left faces need to use align_megaface_failures.m to align
folder = 'E:\datasets\megaface\megafacedata\FlickrFinal2';
if ~exist('image_list','var')
list_file = 'E:\datasets\megaface\devkit\templatelists\megaface_features_list.json_10000_1';
json_string = fileread(list_file);
image_list = regexp(json_string(8:end), '"(.*?)"','tokens');
for i=1:length(image_list)
image_list{i} = [folder '/' image_list{i}{1}];
end;
end;
target_folder = 'F:\datasets\MegaFace\Megaface\aligned';
if exist(target_folder, 'dir')==0
mkdir(target_folder);
end;
pdollar_toolbox_path='D:/face project/pdollar-toolbox';
addpath(genpath(pdollar_toolbox_path));
MTCNN_path = 'D:\face project\MTCNN_face_detection_alignment\code\codes\MTCNNv2';
caffe_model_path=[MTCNN_path , '/model/'];
addpath(genpath(MTCNN_path));
gpu_id = 0;
MatMTCNN('init_model', caffe_model_path, gpu_id);
coord5points = [30.2946, 65.5318, 48.0252, 33.5493, 62.7299; ...
51.6963, 51.5014, 71.7366, 92.3655, 92.2041];
imgSize = [112, 96];
%three steps's threshold
threshold=[0.6 0.7 0.7]
MatMTCNN('set_threshold', threshold);
minsize = 100;
%scale factor
factor=0.709;
%load caffe models
prototxt_dir =strcat(caffe_model_path,'/det1.prototxt');
model_dir = strcat(caffe_model_path,'/det1.caffemodel');
PNet=caffe.Net(prototxt_dir,model_dir,'test');
prototxt_dir = strcat(caffe_model_path,'/det2.prototxt');
model_dir = strcat(caffe_model_path,'/det2.caffemodel');
RNet=caffe.Net(prototxt_dir,model_dir,'test');
prototxt_dir = strcat(caffe_model_path,'/det3.prototxt');
model_dir = strcat(caffe_model_path,'/det3.caffemodel');
ONet=caffe.Net(prototxt_dir,model_dir,'test');
prototxt_dir = strcat(caffe_model_path,'/det4.prototxt');
model_dir = strcat(caffe_model_path,'/det4.caffemodel');
LNet=caffe.Net(prototxt_dir,model_dir,'test');
faces=cell(0);
image_list_len = length(image_list);
landmark_list = cell(image_list_len,1);
for image_id = 1:image_list_len
clear cropImg;
[file_folder, file_name, file_ext] = fileparts(image_list{image_id});
target_filename = strrep(image_list{image_id},folder, target_folder);
if ~exist([image_list{image_id}, '.json'],'file')
continue;
end;
if exist(target_filename, 'file')
continue;
end;
try
img = imread(image_list{image_id});
catch
continue;
end;
if size(img, 3) < 3
img(:,:,2) = img(:,:,1);
img(:,:,3) = img(:,:,1);
end
% result1 = MatMTCNN('detect', img, min(imgSize(1), imgSize(2)));
% numbox1=size(result1.bounding_box,1);
% original_img = img;
assert(strcmp(target_filename, image_list{image_id})==0);
[file_folder, file_name, file_ext] = fileparts(target_filename);
if exist(file_folder,'dir')==0
mkdir(file_folder);
end;
json = parse_json(fileread([image_list{image_id} '.json']));
if isfield(json{1}, 'landmarks')
landmarks = json{1}.landmarks;
landmarks_fields = fieldnames(json{1}.landmarks);
if length(landmarks_fields) == 3
try
facial3points = [json{1}.landmarks.n1.x json{1}.landmarks.n0.x json{1}.landmarks.n2.x;
json{1}.landmarks.n1.y json{1}.landmarks.n0.y json{1}.landmarks.n2.y];
catch
continue;
end;
Tfm = cp2tform(facial3points', coord5points(:,1:3)'*1.5+repmat(imgSize*0.25,[3 1]), 'similarity');
img = imtransform(img, Tfm, 'XData', [1 imgSize(1)*2],...
'YData', [1 imgSize(1)*2], 'Size', [imgSize(1)*2 imgSize(1)*2]);
result = MatMTCNN('detect', img, imgSize(1));
default_face = 1;
if ~isempty(result.bounding_box)
facial5points = [result.points(default_face,1:2:9);result.points(default_face,2:2:10)];
Tfm = cp2tform(facial5points', coord5points', 'similarity');
cropImg = imtransform(img, Tfm, 'XData', [1 imgSize(2)],...
'YData', [1 imgSize(1)], 'Size', imgSize);
fprintf('%d/%d, %s [%d,%d]\n',image_id,image_list_len,target_filename,int32(result.bounding_box(default_face,3)),int32(result.bounding_box(default_face,4)));
else
continue;
end;
end;
end;
if exist('cropImg','var')
imwrite(cropImg, target_filename);
end;
% show detection result
% if numbox1 > 0 && ~exist('cropImg','var')
% figure(3);
% imshow(original_img);
% hold on;
% if ~isempty(result1.bounding_box)
% for j=1:numbox1
% rectangle('Position',result1.bounding_box(j,:),'Edgecolor','r','LineWidth',3); % Note that the predicted bbox could still not close to labeled bbox
% plot(result1.points(j,1:2:9),result1.points(j,2:2:10),'g.','MarkerSize',10);
% end;
% end;
% hold off;
% pause
% end;
% if numbox1 ==0 && exist('cropImg','var')
% numbox=size(result.bounding_box,1);
% figure(1);
% imshow(img);
% hold on;
% if ~isempty(result.bounding_box)
% for j=1:numbox
% rectangle('Position',result.bounding_box(default_face,:),'Edgecolor','r','LineWidth',3); % Note that the predicted bbox could still not close to labeled bbox
% plot(result.points(default_face,1:2:9),result.points(default_face,2:2:10),'g.','MarkerSize',10);
% end;
% end;
% hold off;
% figure(2);
% imshow(cropImg);
% figure(3);
% imshow(original_img);
% hold on;
% if ~isempty(result1.bounding_box)
% for j=1:numbox1
% rectangle('Position',result1.bounding_box(j,:),'Edgecolor','r','LineWidth',3); % Note that the predicted bbox could still not close to labeled bbox
% plot(result1.points(j,1:2:9),result1.points(j,2:2:10),'g.','MarkerSize',10);
% end;
% end;
% hold off;
% pause
%
% end;
end;