forked from woshiwwwppp/ryolov3research-pytorch-master
-
Notifications
You must be signed in to change notification settings - Fork 0
/
generate_data.py
206 lines (191 loc) · 6.69 KB
/
generate_data.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
# -*- coding: utf-8 -*-
import random
import xml.etree.ElementTree as ET
import os
from collections import Counter
# from kmeans import YOLO_Kmeans
import math
import cv2
import numpy as np
f = open('object_classes_all.txt','r')
cla = f.readlines()
pre_classes =[c.strip().split(' ')[0] for c in cla]
classes_name = []
print(pre_classes)
train_percent = 0.85 # 20% 用来验证 80% 用来训练
txtsavepath = 'train_data'
datapath = r'H:\pyProject\detection\rbox\kaggle-airbus-ship-detection-challenge-master\dataset\data'
filedir = os.listdir(datapath)
picDir="H:\\pyProject\\detection\\rbox\\kaggle-airbus-ship-detection-challenge-master\\dataset\\data\\images\\"
ftest = open('train_data\\test_v2.txt', 'w')
ftrain = open('train_data\\train_v2.txt', 'w')
not_enough = []
#数据问题标记,0为正常,1为不正常
a = 0
train_ = 0
test_ = 0
for filedir in filedir:
total_files = os.listdir(os.path.join(datapath, filedir))
random.shuffle(total_files)
total_xml = []
total_image = []
#找到所有xml,image
for file in total_files:
filetype = os.path.splitext(file)[1]
if filetype == '.xml':
total_xml.append(file)
per = ET.parse(os.path.join(datapath, filedir,file))
root = per.getroot()
for Object in root.findall('object'):
name = Object.find('name').text
if name in pre_classes:
classes_name.append(name)
else:
print(name,'in',file,'is not in defult classes!!!')
a +=1
# if name == 'd':
# Object.find('name').text = 'D'
# per.write(os.path.join(datapath, filedir,file))
else:
total_image.append(file)
#确认图片与xml对应
for image in total_image:
if image.split('.')[0] +'.xml' not in total_xml:
print(image,'has no xml file!!!')
a += 1
for xml in total_xml:
if xml.split('.')[0] +'.jpg' not in total_image:
print(xml,'has no jpg file!!!')
a+=1
random.shuffle(total_xml)
num = len(total_xml)
list = range(num)
trainNum = int(num * train_percent)
train = random.sample(list, trainNum)
for i in list:
name = os.path.join(datapath, filedir, total_xml[i])+ '\n'
if i in train:
ftrain.write(name)
train_ +=1
else:
ftest.write(name)
test_ +=1
print(filedir,int(len(total_files)/2),train_,test_)
ftrain.close()
ftest.close()
clas = Counter(classes_name)
clakey = [c for c in clas.keys()]
clav = [c for c in clas.values()]
classes_1 = []
not_enough_class = []
classes = []
for i in range(len(clakey)):
if clav[i] <100:
not_enough_class.append(clakey[i])
else:
classes_1.append(clakey[i])
for c in pre_classes:
if c in classes_1:
classes.append(c)
print(clas)
print('not_enough_file:',not_enough)
print('not_enough_class:',not_enough_class)
print('classes:',classes,len(classes))
print('total train:',train_)
print('total test:',test_)
if a == 0:
print('data checked!')
else:
print('data error!')
#输出模型需要的类别目录
co=open('train_data/train_classes.txt','w')
for c in classes:
co.write(c)
co.write('\n')
co.close()
sets=[ 'train_v2','test_v2','val']
#提取训练和测试数据
def convert_annotation(image_path, list_file,image_set):
img_h=768.
img_w=768.
picOut="H:\\pyProject\\detection\\rbox\\kaggle-airbus-ship-detection-challenge-master\\dataset\\%s\\"%(image_set)
in_file = open(image_path,encoding='UTF-8')
tree=ET.parse(in_file)
root = tree.getroot()
filename = root.find('filename').text
picture = cv2.imread(picDir + filename)
height, width = picture.shape[:2]
changeA = width / img_h
rePic = cv2.resize(picture, (int(width/changeA), int(height/changeA)))
reheight, rewidth = rePic.shape[:2]
pic=np.zeros((int(img_h),int(img_w),3),dtype=np.uint8)
startY=img_h/2-reheight/2
pic[int(startY):int(startY)+reheight,:,:]=rePic
imageSave =picOut+filename
cv2.imwrite(imageSave, pic)
for obj in root.iter('object'):
difficult = obj.find('difficult').text
cls = obj.find('name').text
if cls not in classes or int(difficult)==1:
continue
cls_id = classes.index(cls)
xmlbox = obj.find('bndbox')
cx = (float(xmlbox.find('x').text))/changeA
cy = (float(xmlbox.find('y').text))/changeA
rw = (float(xmlbox.find('w').text))/changeA
rh = (float(xmlbox.find('h').text))/changeA
angle = float(xmlbox.find('angle').text)
#角度转弧度
angle=angle/180.0*math.pi
if rh<rw:
angle=angle+math.pi/2
temp=rw
rw=rh
rh=temp
if angle>0:
while(angle>math.pi):
angle-=math.pi
else :
while(angle<=0):
angle+=math.pi
# bow_x = b[0] + b[2] / 2 * math.cos(float(b[4]))
# bow_y = b[1] - b[2] / 2 * math.sin(float(b[4]))
#
# tail_x = b[0] - b[2] / 2 * math.cos(float(b[4]))
# tail_y = b[1] + b[2] / 2 * math.sin(float(b[4]))
#
# # print(bow_x,bow_y,tail_x,tail_y)
#
# x1 = round(bow_x + b[3] / 2 * math.sin(float(b[4])))
# y1 = round(bow_y + b[3] / 2 * math.cos(float(b[4])))
#
# x2 = round(tail_x + b[3] / 2 * math.sin(float(b[4])))
# y2 = round(tail_y + b[3] / 2 * math.cos(float(b[4])))
#
# x3 = round(tail_x - b[3] / 2 * math.sin(float(b[4])))
# y3 = round(tail_y - b[3] / 2 * math.cos(float(b[4])))
#
# x4 = round(bow_x - b[3] / 2 * math.sin(float(b[4])))
# y4 = round(bow_y - b[3] / 2 * math.cos(float(b[4])))
#
# # print(bow_x,bow_y,tail_x,tail_y)
# print(x1, y1, x2, y2, x3, y3, x4, y4)
# list_file.write(" " + "%d %d %d %d %d %d %d %d"%(x1,y1,x2,y2,x3,y3,x4,y4) + ' ' + str(cls_id))
list_file.write(filename+",%f,%f,%f,%f,%f\n" % (cx,cy,rw,rh,angle))
for image_set in sets:
image_ids = open('train_data\\%s.txt'%(image_set),encoding='UTF-8').read().strip().split()
list_file = open('data_%s.txt'%(image_set), 'w',encoding='UTF-8')
list_file.write('ImageID,x,y,height,width,rotate\n')
for image_id in image_ids:
image_path = image_id.split('.')[0] +'.jpg'
convert_annotation(image_id, list_file,image_set)
# try:
# convert_annotation(image_xml, list_file)
# except:
# print(image_xml,'is broken')
list_file.close()
cluster_number = 9
filename = "data_train.txt"
# kmeans = YOLO_Kmeans(cluster_number, filename)
# kmeans.txt2clusters()
#