-
Notifications
You must be signed in to change notification settings - Fork 3
/
processes.py
264 lines (236 loc) · 12.2 KB
/
processes.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
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
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
import os
import cv2
import numpy as np
from PIL import Image, ImageDraw, ImageFont
import shutil
import copy
import time
seq = os.sep
pwd = os.path.abspath(r'.%s'%seq)
def write_train_val(file, stage, name):
with open(os.path.join(pwd, 'data', stage, '%s.txt'%name), 'w') as obj:
for i in file:
obj.write(i)
obj.write('\n')
chinesechar = {'皖': 0, '沪': 1, '津': 2, '渝': 3, '冀': 4, '晋': 5, '蒙': 6, '辽': 7, '吉': 8, '黑': 9, '苏': 10, '浙': 11, \
'京': 12, '闽': 13, '赣': 14, '鲁': 15, '豫': 16, '鄂': 17, '湘': 18, '粤': 19, '桂': 20, '琼': 21, '川': 22, \
'贵': 23, '云': 24, '西': 25, '陕': 26, '甘': 27, '青': 28, '宁': 29, '新': 30}
chinesechar_reverse = {0: '皖', 1: '沪', 2: '津', 3: '渝', 4: '冀', 5: '晋', 6: '蒙', 7: '辽', 8: '吉', 9: '黑', 10: '苏', \
11: '浙', 12: '京', 13: '闽', 14: '赣', 15: '鲁', 16: '豫', 17: '鄂', 18: '湘', 19: '粤', 20: '桂', \
21: '琼', 22: '川', 23: '贵', 24: '云', 25: '西', 26: '陕', 27: '甘', 28: '青', 29: '宁', 30: '新'}
alphaenglish = {'a': 0, 'b': 1, 'c': 2, 'd': 3, 'e': 4, 'f': 5, 'g': 6, 'h': 7, 'j': 8, 'k': 9, 'l': 10, 'm': 11, 'n': 12, \
'p': 13, 'q': 14, 'r': 15, 's': 16, 't': 17, 'u': 18, 'v': 19, 'w': 20, 'x': 21, 'y': 22, 'z': 23, '0': 24, \
'1': 25, '2': 26, '3': 27, '4': 28, '5': 29, '6': 30, '7': 31, '8': 32, '9': 33}
alphaenglish_reverse = {0: 'a', 1: 'b', 2: 'c', 3: 'd', 4: 'e', 5: 'f', 6: 'g', 7: 'h', 8: 'j', 9: 'k', 10: 'l', 11: 'm', \
12: 'n', 13: 'p', 14: 'q', 15: 'r', 16: 's', 17: 't', 18: 'u', 19: 'v', 20: 'w', 21: 'x', 22: 'y', \
23: 'z', 24: '0', 25: '1', 26: '2', 27: '3', 28: '4', 29: '5', 30: '6', 31: '7', 32: '8', 33: '9'}
classes = ['皖', '沪', '津', '渝', '冀', '晋', '蒙', '辽', '吉', '黑', '苏', '浙', '京', '闽', '赣', '鲁', '豫', '鄂', '湘', \
'粤', '桂', '琼', '川', '贵', '云', '西', '陕', '甘', '青', '宁', '新', 'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', \
'j', 'k', 'l', 'm', 'n', 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', 'x', 'y', 'z', '0', '1', '2', '3', '4', '5', \
'6', '7', '8', '9']
print('len(classes): ', classes.__len__())
def warp(height, width, star_points, src_img):
# 目标尺寸
# height = 35
# width = 110
pts1 = np.float32([star_points[0], star_points[1], star_points[2], star_points[3]])
pts2 = np.float32([[0, 0], [width, 0], [0, height], [width, height]])
M = cv2.getPerspectiveTransform(pts1, pts2)
dst = cv2.warpPerspective(src_img, M, (width, height))
dst = cv2.cvtColor(dst, cv2.COLOR_BGR2RGB)
# cv2.imshow('img', dst)
# cv2.waitKey(0)
# cv2.destroyWindow('img')
return dst
def cal_txt(shape, xmin, ymin, xmax, ymax, name):
h, w, c = shape
center_x = (xmin+xmax)/2
center_y = (ymin+ymax)/2
center_x = center_x/w
center_y = center_y/h
item_h = (ymax-ymin)/h
item_w = (xmax-xmin)/w
cor = np.array([center_x, center_y, item_w, item_h])
if (np.any(cor > 1)) or (np.any(cor < 0)):
return -1, -1, -1, -1
return center_x,center_y,item_w,item_h
def cvshow(img, xmin, ymin ,xmax, ymax, rdc_x,rdc_y, ldc_x, ldc_y, ruc_x, ruc_y, luc_x, luc_y, platestring):
cv2.rectangle(img, (xmin, ymin), (xmax, ymax), [255, 0, 0], 1)
cv2.circle(img, (rdc_x, rdc_y), 3, [255, 0, 255], 1)
cv2.circle(img, (ldc_x, ldc_y), 3, [255, 0, 255], 1)
cv2.circle(img, (luc_x, luc_y), 3, [255, 0, 255], 1)
cv2.circle(img, (ruc_x, ruc_y), 3, [255, 0, 255], 1)
img_PIL = Image.fromarray(img).convert('RGB')
font = ImageFont.truetype('simsun.ttc', 50)
# 字体颜色
fillColor = (255, 0, 0)
draw = ImageDraw.Draw(img_PIL)
draw.text((xmin, ymin-39 ,), platestring, font=font, fill=fillColor)
img = np.asarray(img_PIL)
cv2.imshow('i', img)
cv2.waitKey(0)
cv2.destroyWindow('i')
def rectangle_save(img, xmin, ymin, xmax, ymax, rdc_x, rdc_y, ldc_x, ldc_y, ruc_x, \
ruc_y, luc_x, luc_y, rectangle, ori_i,rectangle_label):
rectangle_img = img[ymin:ymax + 1, xmin:xmax + 1, :]
h, w, c = rectangle_img.shape
luc_x, luc_y = (luc_x-xmin)/w, (luc_y-ymin)/h
ldc_x, ldc_y = (ldc_x-xmin)/w, (ldc_y-ymin)/h
rdc_x, rdc_y = (rdc_x-xmin)/w, (rdc_y-ymin)/h
ruc_x, ruc_y = (ruc_x-xmin)/w, (ruc_y-ymin)/h
#
# cv2.circle(rectangle_img, (int(rdc_x*w), int(rdc_y*h)), 3, [255, 0, 255], 1)
# cv2.circle(rectangle_img, (int(ldc_x*w), int(ldc_y*h)), 3, [255, 0, 255], 1)
# cv2.imshow('i', rectangle_img)
# cv2.waitKey(0)
# cv2.destroyWindow('i')
corner = [luc_x, luc_y, ldc_x, ldc_y, rdc_x, rdc_y, ruc_x, ruc_y]
cor = np.array(corner)
if (np.any(cor>1)) or (np.any(cor<0)):
return -1
# rectangle_img = img[ymin:ymax + 1, xmin:xmax + 1, :]
cv2.imwrite(os.path.join(rectangle, ori_i), rectangle_img)
with open(os.path.join(rectangle_label, ori_i.replace('.jpg','.txt')), 'w') as obj:
obj.write(' '.join([str(a) for a in corner]))
pwd = os.path.abspath('./')
path = os.path.join(pwd, 'data', 'CCPD_part')
widths = []
heights = []
# for root_dir, dir, file in os.walk(path):
# if len(file)==0:
# continue
# if 'base' not in root_dir:
# continue
# count=0
# for i in file:
# count+=1
# if count%999==0:
# print(count)
# # img = cv2.imread(os.path.join(root_dir, i))
# i = i.replace('.jpg', '').replace('.png', '').replace('&', '_').replace('-', '_').split('_')
# angle = [int(i[1]), int(i[2])]
# xmin, ymin, xmax, ymax = int(i[3]), int(i[4]), int(i[5]), int(i[6])
# rdc_x, rdc_y, ldc_x, ldc_y, luc_x, luc_y, ruc_x, ruc_y = int(i[7]), int(i[8]), int(i[9]), int(i[10]), \
# int(i[11]), int(i[12]), int(i[13]), int(i[14])
#
# # shape = img.shape
# # start_points = [[luc_x, luc_y], [ruc_x, ruc_y], [ldc_x, ldc_y], [rdc_x, rdc_y]]
# # warp(70, 160, start_points, img)
#
# width = (np.sqrt(np.square(rdc_x-ldc_x)+np.square(rdc_y-ldc_y))+\
# np.sqrt(np.square(ruc_x-luc_x)+np.square(ruc_y-luc_y)))/2
# height = (np.sqrt(np.square(rdc_x-ruc_x)+np.square(rdc_y-ruc_y))+\
# np.sqrt(np.square(ldc_x-luc_x)+np.square(ldc_y-luc_y)))/2
# if (abs(np.mean(widths)- width) > 10) or (abs(np.mean(heights) - height) > 10):
# continue
# # if (width > 300) or (height > 119):
# # print(root_dir, i)
# # # cvshow(img, xmin, ymin, xmax, ymax, rdc_x, rdc_y, ldc_x, ldc_y, ruc_x, ruc_y, luc_x, luc_y, '111111')
# # continue
# widths.append(width)
# heights.append(height)
mean_width = 190 # np.mean(widths)
mean_height = 60 # np.mean(height)
print('车牌的平均宽度为:', mean_width)
print('车牌的平均高度为:', mean_height)
stage1_images = os.path.join(pwd, 'data', 'stage1', 'images')
stage1_labels = os.path.join(pwd, 'data', 'stage1', 'labels')
stage1_test = os.path.join(pwd, 'data', 'stage1', 'test')
rectangle = os.path.join(pwd, 'data', 'corner', 'rectangle')
rectangle_label = os.path.join(pwd, 'data', 'corner', 'label')
rectangle_test = os.path.join(pwd, 'data', 'corner', 'test')
# labeltxt = os.path.join(pwd, 'data', 'crnn', 'labeltxt')
# warpimg = os.path.join(pwd, 'data', 'crnn', 'warpimg')
# warpimg_test = os.path.join(pwd, 'data', 'crnn', 'test')
dirs = [stage1_images, stage1_labels, rectangle, rectangle_label, stage1_test, rectangle_test]
delete = input('是否删除之前的文件夹?Y or N')
for i in dirs:
if delete.upper()=='Y':
try:
shutil.rmtree(i)
except:
pass
os.makedirs(i)
print('begin proprecess...')
start = time.time()
count = 0
for root_dir, dir, file in os.walk(path):
if len(file)==0:
continue
if ('&' not in file[0]) or ('-' not in file[0]):
continue
np.random.shuffle(file)
# file = file[:30000]
for i in file:
# if count > 1000:
# break
count+=1
ori_i = copy.deepcopy(i)
name = copy.deepcopy(i)
img = cv2.imread(os.path.join(root_dir, i))
shape = img.shape
i = i.replace('.jpg', '').replace('.png', '').replace('&', '_').replace('-', '_').split('_')
angle = [int(i[1]), int(i[2])]
xmin, ymin, xmax, ymax = int(i[3]), int(i[4]), int(i[5]), int(i[6])
#stage1
convert = cal_txt(shape, xmin, ymin, xmax, ymax, i)
if convert[0] < 0:
continue
name = name.replace('.jpg', '.txt').replace('.png', '.txt')
if count<30000:
shutil.copyfile(os.path.join(root_dir, ori_i), os.path.join(pwd, 'data', 'stage1', 'images', ori_i))
with open(os.path.join(pwd, 'data', 'stage1', 'labels', name.replace('.jpg', '.txt')), 'w') as obj:
obj.write("0 " + " ".join([str(a) for a in convert]) + '\n')
rdc_x, rdc_y, ldc_x, ldc_y, luc_x, luc_y, ruc_x, ruc_y = int(i[7]), int(i[8]), int(i[9]), int(i[10]), \
int(i[11]), int(i[12]), int(i[13]), int(i[14])
xmax = np.max([rdc_x, ruc_x, xmax])
xmin = np.min([luc_x, ldc_x, xmin])
ymin = np.min([luc_y, ruc_y, ymin])
ymax = np.max([ldc_y, rdc_y, ymax])
platestring = chinesechar_reverse[int(i[15])]
for j in i[16:-2]:
platestring += alphaenglish_reverse[int(j)]
platestring = platestring.upper()
# print(platestring.upper())
# cvshow(img, xmin, ymin, xmax, ymax, rdc_x, rdc_y, ldc_x, ldc_y, ruc_x, ruc_y, luc_x, luc_y, platestring)
start_points = [[luc_x, luc_y], [ruc_x, ruc_y], [ldc_x, ldc_y], [rdc_x, rdc_y]]
# warp_img = warp(mean_height, mean_width, start_points, img)
rectangle_save(img, xmin, ymin, xmax, ymax, rdc_x, rdc_y, ldc_x, ldc_y, ruc_x, \
ruc_y, luc_x, luc_y, rectangle, ori_i, rectangle_label)
# warp_img = Image.fromarray(warp_img)
# warp_img.save(os.path.join(warpimg, platestring+'.jpg'))
# with open(os.path.join(labeltxt, ori_i.replace('.jpg', '.txt').replace('.png', '.txt')), 'w') as obj:
# obj.write(platestring)
#自动划分训练集和验证集
namelist = [os.path.join(stage1_images, i) for i in os.listdir(stage1_images)]
trainchoose = list(np.random.choice(namelist, len(namelist)*9//10, replace=False))
valchoose = list(set(namelist) - set(trainchoose))
testchoose = list(np.random.choice(valchoose, len(valchoose)*9//10, replace=False))
valchoose = list(set(valchoose) - set(testchoose))
for i in testchoose:
shutil.move(i, i.replace('images', 'test'))
shutil.move(i.replace('images', 'labels').replace('.jpg', '.txt'), \
i.replace('.jpg', '.txt').replace('images', 'test'))
write_train_val(testchoose, 'stage1', 'test')
print('训练集图片个数:{},验证集图片个数:{}'.format(len(trainchoose), len(valchoose)))
write_train_val(trainchoose, 'stage1', 'train')
write_train_val(valchoose, 'stage1', 'valid')
namelist = [os.path.join(rectangle, i) for i in os.listdir(rectangle)]
trainchoose = list(np.random.choice(namelist, len(namelist)*9//10, replace=False))
valchoose = list(set(namelist) - set(trainchoose))
testchoose = list(np.random.choice(valchoose, len(valchoose)*9//10, replace=False))
valchoose = list(set(valchoose) - set(testchoose))
for i in testchoose:
shutil.move(i, i.replace('rectangle', 'test'))
shutil.move(i.replace('rectangle', 'label').replace('.jpg', '.txt'), \
i.replace('.jpg', '.txt').replace('rectangle', 'test'))
# namelist = [os.path.join(warpimg, i) for i in os.listdir(warpimg)]
# trainchoose = list(np.random.choice(namelist, len(namelist)*9//10, replace=False))
# valchoose = list(set(namelist) - set(trainchoose))
# testchoose = list(np.random.choice(valchoose, len(valchoose)*9//10, replace=False))
# valchoose = list(set(valchoose) - set(testchoose))
# for i in testchoose:
# shutil.move(i, i.replace('warpimg', 'test'))
# shutil.move(i.replace('warpimg', 'labeltxt').replace('.jpg', '.txt'), \
# i.replace('.jpg', '.txt').replace('warpimg', 'test'))
print('use: ', round((time.time() - start)/60, 3), 'min')