-
Notifications
You must be signed in to change notification settings - Fork 1
/
lm_demo.m
120 lines (100 loc) · 4.11 KB
/
lm_demo.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
%% demo code for facial landmark detection
close all
clear
clc
addpath('/user/HS204/m09113/facer2vm_project_area/people/Philipp/Zhenhua_lms_DAC_CSR/')
run /user/HS204/m09113/facer2vm_project_area/Share/DAC_CSR/vlfeat-0.9.20/toolbox/vl_setup.m;
%img_list = dir('/user/HS204/m09113/facer2vm_project_area/Share/DAC_CSR/data/*.png');
fid = fopen('/user/HS204/m09113/facer2vm_project_area/data/300VW_Dataset_2015_12_14/bb_clicked_philipp.log','r');
C = textscan(fid, '%s %d %d %d %d', 'Delimiter', ' ');
first_img_list = C{1};
%class(img_list)
bboxs = double(cell2mat(C(1:end,2:5)));
bboxs (1:end,3) = bboxs(1:end,3)-bboxs(1:end,1);
bboxs (1:end,4) = bboxs(1:end,4)-bboxs(1:end,2);
load /user/HS204/m09113/facer2vm_project_area/Share/DAC_CSR/cr_model_68.mat;
show_results = false;
write_pts = true;
%global frame_num
faceDetector = vision.CascadeObjectDetector();
disp('created facedetector')
parfor i = 1:length(first_img_list)
pre_bb = [ 0 0 0 0 ];
% load image
%img = imread(['/user/HS204/m09113/facer2vm_project_area/Share/DAC_CSR/data/', img_list(i).name]);
%class(img_list{i})
first_img_list{i}(1:end-10)
all_vid_frames = dir([first_img_list{i}(1:end-10) '*.png']);
if write_pts
mkdir ([first_img_list{i}(1:end-17) 'CSR_lms']);
end
for frame_num = 1:length(all_vid_frames)
disp(all_vid_frames(frame_num).name);
%try
%img = imread(first_img_list{i});
img = imread([first_img_list{i}(1:end-10) all_vid_frames(frame_num).name]);
if frame_num==1
bbox = bboxs(i,1:end);
else
if pre_bb(3) > 20 && pre_bb(4) > 20
bbox = pre_bb;
else
% Create a cascade detector object.
bbox = step(faceDetector, img);
if size(bbox,1) > 0
bbox = bbox(1,1:end);
else
bbox = pre_bb;
end
end
end
% load bbox and ground truth facial landmarks
%load(['/user/HS204/m09113/facer2vm_project_area/Share/DAC_CSR/data/', img_list(i).name(1:end-3), 'mat']);
%gt_lmk = lmk_bbox.gt_lmk;
%bbox = lmk_bbox.bbox;
% landmark initalisation
init_lmk = project_s2b(cr_model.mean_shape, bbox);
%if frame_num > 580
% init_lmk
%end
% face landmarking
pre_lmk = fit_sdt(rgb2gray(img), init_lmk, cr_model);
% calc bb around predicted lms, that can be used as init for next
% frame
pre_bb = [ 0 0 0 0 ];
pre_bb(1)=min(pre_lmk(1:end/2));
pre_bb(2)=min(pre_lmk(end/2+1:end));
pre_bb(3)=max(pre_lmk(1:end/2)) - pre_bb(1);
pre_bb(4)=max(pre_lmk(end/2+1:end)) - pre_bb(2);
% display result
if show_results && frame_num>550
imshow(img);
hold on;
%plot(gt_lmk(1:end/2), gt_lmk(end/2+1:end), 'yo', 'markerfacecolor', 'y');
plot(pre_lmk(1:end/2), pre_lmk(end/2+1:end), 'ro', 'markerfacecolor', 'r');
hold off;
title('press any key for the next image');
pause();
end
% write lm files
if write_pts
ofid = fopen([first_img_list{i}(1:end-17) 'CSR_lms/' all_vid_frames(frame_num).name(1:end-3) 'pts'], 'w');
fprintf(ofid,'version: 1\n');
fprintf(ofid,'n_points: 68\n');
fprintf(ofid,'{\n');
num_lms=length(pre_lmk(1:end/2));
for lmi = 1:num_lms
fprintf(ofid,'%f %f\n',pre_lmk(lmi),pre_lmk(lmi+num_lms));
end
fprintf(ofid,'}\n');
fclose(ofid);
end
%catch
% err = lasterror;
% disp(err);
% disp(err.message);
% disp(err.stack);
% disp(err.identifier);
%end
end
end