-
Notifications
You must be signed in to change notification settings - Fork 7
/
Copy pathmake_dataset.py
197 lines (170 loc) · 6.21 KB
/
make_dataset.py
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
# -*- coding:UTF-8 -*-
import sys
import cv2
import six
import os
import math
import time
import numpy as np
from tqdm import tqdm
# 计算两个框的iou
def compute_iou(rec1, rec2):
"""
computing IoU
:param rec1: (y0, x0, y1, x1), which reflects
(top, left, bottom, right)
:param rec2: (y0, x0, y1, x1)
:return: scala value of IoU
"""
# computing area of each rectangles
S_rec1 = (rec1[2] - rec1[0]) * (rec1[3] - rec1[1])
S_rec2 = (rec2[2] - rec2[0]) * (rec2[3] - rec2[1])
# computing the sum_area
sum_area = S_rec1 + S_rec2
# find the each edge of intersect rectangle
left_line = max(rec1[1], rec2[1])
right_line = min(rec1[3], rec2[3])
top_line = max(rec1[0], rec2[0])
bottom_line = min(rec1[2], rec2[2])
# judge if there is an intersect
if left_line >= right_line or top_line >= bottom_line:
return 0
else:
intersect = (right_line - left_line) * (bottom_line - top_line)
return intersect / (sum_area - intersect)
# 将框整理为正方形
def get_square_box(box):
"""Get the square boxes which are ready for CNN from the boxes"""
left_x = box[0]
top_y = box[1]
right_x = box[2]
bottom_y = box[3]
box_width = right_x - left_x
box_height = bottom_y - top_y
# Check if box is already a square. If not, make it a square.
diff = box_height - box_width
delta = int(abs(diff) / 2)
if diff == 0: # Already a square.
return box
elif diff > 0: # Height > width, a slim box.
left_x -= delta
right_x += delta
if diff % 2 == 1:
right_x += 1
else: # Width > height, a short box.
top_y -= delta
bottom_y += delta
if diff % 2 == 1:
bottom_y += 1
# Make sure box is always square.
assert ((right_x - left_x) == (bottom_y - top_y)), 'Box is not square.'
return left_x, top_y, right_x, bottom_y
# 读取标注
def read_points(file_name=None):
"""
Read points from .pts file.
"""
points = []
with open(file_name) as file:
line_count = 0
for line in file:
if "version" in line or "points" in line or "{" in line or "}" in line:
continue
else:
loc_x, loc_y = line.strip().split(sep=" ")
points.append([float(loc_x), float(loc_y)])
line_count += 1
return points
# 绘制关键点
def draw_landmark_point(image, points):
"""
Draw landmark point on image.
"""
for point in points:
cv2.circle(image, (int(point[0]), int(
point[1])), 1, (0, 255, 0), -1, cv2.LINE_AA)
def make_datasets(image,gray,points):
# 探测图片中的人脸
faces = face_cascade.detectMultiScale(
gray,
scaleFactor = 1.5,
minNeighbors = 5,
minSize = (5,5)
)
#print ("发现{0}个人脸!".format(len(faces)))
#print(faces)
points_translate = np.array(points).T
xmin = min(points_translate[0]).astype(np.int32)
xmax = max(points_translate[0]).astype(np.int32)
ymin = min(points_translate[1]).astype(np.int32)
ymax = max(points_translate[1]).astype(np.int32)
box = [0,0,0,0]
minimal_box = [xmin,ymin,xmax,ymax]
for(x,y,w,h) in faces:
box[0] = x
box[1] = y
box[2] = x + w
box[3] = y + h
iou = compute_iou(np.array(box),np.array(minimal_box))
#print(iou)
if iou > 0.1:
#cv2.rectangle(image,(box[0],box[1]),(box[2],box[3]),(0,255,128),3)
if minimal_box[0] > box[0]:
minimal_box[0] = box[0]
if minimal_box[1] > box[1]:
minimal_box[1] = box[1]
if minimal_box[2] < box[2]:
minimal_box[2] = box[2]
if minimal_box[3] < box[3]:
minimal_box[3] = box[3]
w = minimal_box[2] - minimal_box[0]
h = minimal_box[3] - minimal_box[1]
#print("w,h = ",w,h)
xmin,ymin,xmax,ymax = get_square_box(minimal_box)
points_translate[0] = points_translate[0] - xmin
points_translate[1] = points_translate[1] - ymin
cv2.rectangle(image,(xmin,ymin),(xmax,ymax),(126,0,25),2)
slip_image = image[ymin:ymax,xmin:xmax]
#draw_landmark_point(image,points)
wh = xmax - xmin
slip_image = cv2.resize(slip_image, (224,224), interpolation=cv2.INTER_CUBIC)
slip_image_wh = slip_image.shape[0]
xishu = slip_image_wh/wh
#print(wh)
#print(slip_image_wh)
#print(xishu)
points_translate = points_translate*xishu
#draw_landmark_point(slip_image,points_translate.T)
#cv2.imshow("Find Faces!",image)
#cv2.imshow("slip_image!",slip_image)
#cv2.waitKey(0)
return slip_image,points_translate.T
image_names = []
for image_name in os.listdir("data/data7426/300w/01_Indoor"):
if image_name.split('.')[-1] in ['png']:
image_names.append("data/data7426/300w/01_Indoor/"+image_name)
for image_name in os.listdir("data/data7426/300w/02_Outdoor"):
if image_name.split('.')[-1] in ['png']:
image_names.append("data/data7426/300w/02_Outdoor/"+image_name)
print(image_names)
# 获取训练好的人脸的参数数据，这里直接从GitHub上使用默认值
face_cascade = cv2.CascadeClassifier(r'work/Face-Localization/haarcascade_frontalface_default.xml')
for image_name in image_names:
# 待检测的图片路径
points_name = image_name.replace('.png', '.pts')
points = read_points(points_name)
print(image_name)
# 读取图片
image = cv2.imread(image_name)
gray = cv2.cvtColor(image,cv2.COLOR_BGR2GRAY)
result_image,result_points = make_datasets(image,gray,points)
result_image_name = image_name.split('door/')[1]
cv2.imwrite("data/data7426/300w_224x224/"+result_image_name,result_image)
result_points_name = result_image_name.replace('.png', '.pts')
with open("data/data7426/300w_224x224/"+result_points_name, 'w+') as f:
f.write("version: 1"+'\n') #加\n换行显示
f.write("n_points: 68"+'\n') #加\n换行显示
f.write("{"+'\n') #加\n换行显示
for line in result_points:
f.write(str(line[0])+" "+str(line[1])+'\n')
f.write("}"+'\n') #加\n换行显示