SAR/data.py at master · lsabrinax/SAR · GitHub

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
import os
import scipy.io as scio
from PIL import Image
import cv2
import numpy
import fire
import shutil
import xml.etree.cElementTree as ET
# gt_dir = '/home/sabrina/data/text-recognition/SynthText/gt.mat'
# img_dir = '/home/sabrina/data/text-recognition/SynthText/'

# gt = scio.loadmat(gt_dir)

# def get_text(text):
#     texts = []
#     for t in text:
#         texts.extend(t.split())
#     return texts

# num = 0 #记录裁剪的总的图片数
# for ix, imname in enumerate(gt['imnames'][0]):
#     # img = Image.open(img_dir + imname[0])
#     img = cv2.imread(img_dir + imname[0])
#     txts = get_text(gt['txt'][0, ix])
#     wordBB = gt['wordBB'][0, ix]
#     if len(txts) == 1:
#         wordBB = wordBB[:,:,numpy.newaxis]
#     assert len(txts) == wordBB.shape[2], "txt length is not compatible with wordbb length"
#     for i, t in enumerate(txts):
#         wordbb = wordBB[:,:,i]
#         x, y , w, h = cv2.boundingRect(wordbb.T)
#         try:
#             # crop_img = img.crop((x, y, x+w, y+h))
#             crop_img = img[y:y+h,x:x+w]
#         except:
#             print('crop img failed!')
#             continue
#         cropimg_dir = img_dir + 'cropimg/%d'% ((num % 20) + 1)
#         if not os.path.exists(cropimg_dir):
#             os.makedirs(cropimg_dir)
#         imgname = 'word_%d.jpg' % ((num // 20) + 1)
#         # crop_img.save(os.path.join(cropimg_dir, imgname))
#         cv2.imwrite(os.path.join(cropimg_dir, imgname), crop_img)
#         with open(img_dir + 'cropimg/gt_%d.txt'%((num % 20) + 1), 'a') as f:
#             f.write(imgname + ',' + t +'\n')
#         num += 1
#         print(imname, cropimg_dir, imgname, t, 'done!')

def manage_IC15(gt_dir):
    #gt_dir = '/home/sabrina/data/text-recognition-benchmark/IC15/ch4_training_word_images_gt/'
    gt_file = os.path.join(gt_dir, 'gt.txt')
    split_dir = gt_dir + 'split/'
    # if not os.path.exists(gt_split_dir):
    #     os.makedirs(gt_split_dir)
    num = 0#记录成功读取的图片
    with open(gt_file, 'r') as f:
        gts = f.readlines()
        for gt in gts:
            # split_dir = gt_dir + 'split/%d' % ((num % 20) + 1)
            # if not os.path.exists(split_dir):
            #     os.makedirs(split_dir)
            imgname = gt.strip().split(',', 1)[0]
            label = gt.strip().split(',', 1)[1][1:]
            newname = 'word_%d.png' % num
            # gt_name = 'gt_%d.txt' %((num % 20) + 1)
            try:
                shutil.copy(os.path.join(gt_dir, imgname), os.path.join(split_dir, newname))
                with open(os.path.join(split_dir, 'gt.txt'), 'a') as txt:
                    txt.write(newname+','+label+'\n')
            except:
                print('copy failed!')
                continue
            num += 1
            print(imgname, 'done!')

def split_SVT(svt_dir):
    #svt_dir='/home/sabrina/data/text-recognition-benchmark/svt1/'
    xml_file = svt_dir + 'train.xml'
    trees = ET.parse(xml_file)
    num = 0
    for img in trees.iter(tag='image'):
        imgname = img.find('imageName').text
        for rect in img.iter('taggedRectangle'):
            h = int(rect.get('height'))
            w = int(rect.get('width'))
            x = int(rect.get('x'))
            y = int(rect.get('y'))
            word = rect.find('tag').text
            newname = 'word_%d.png' % num
            img_dir = svt_dir + 'split/'
            # if not os.path.exists(img_dir):
            #     os.makedirs(img_dir)
            # gt_dir = svt_dir + 'split/gt/'
            # if not os.path.exists(gt_dir):
            #     os.makedirs(gt_dir)
            # gt_name = 'gt_%d.txt' % ((num % 20) + 1)

            try:
                inputimg = cv2.imread(svt_dir+imgname)
                # print('read img success!')
                cropimg = inputimg[y:y+h,x:x+w]
                # print('cropimg success!')
                cv2.imwrite(img_dir+newname, cropimg)
                # print('write img success!')
            except:
                print('cop img failed')
                continue
            with open(img_dir+'gt.txt', 'a') as f:
                f.write(newname+',"'+word+'"\n')
            num += 1
            print(imgname,'done!')

def split_IIIT5K(iii_dir):
    #iii_dir = '/home/sabrina/data/text-recognition-benchmark/IIIT5K/'
    label_file = iii_dir+'trainCharBound'
    label_data = scio.loadmat(label_file)['trainCharBound'][0]
    num = 0
    for data in label_data:
        imgname = data['ImgName'][0]
        label = data['chars'][0]
        newname = 'word_%d.png' % num
        img_dir = iii_dir+'/split/'
        # if not os.path.exists(img_dir):
        #     os.makedirs(img_dir)
        try:
            shutil.copy(iii_dir+imgname, img_dir+newname)
        except:
            print('copy image failed!')
            continue
        # gt_dir = iii_dir+'split/gt/'
        # if not os.path.exists(gt_dir):
        #     os.makedirs(gt_dir)
        # gt_name = 'gt_%d.txt' % ((num % 20) + 1)
        with open(img_dir+'gt.txt', 'a') as f:
            f.write(newname+',"'+label+'"\n')
        num += 1
        print(imgname,'done!')


def split_Synth90K(syn_dir):
    #90k_dir = 'home/sabrina/data/text-recognition/Synth90K/Synth90K/'
    gt_file = syn_dir + 'annotation_train.txt'
    num = 0
    with open(gt_file, 'r') as txt:
        for line in txt.readlines():
            imgname = line.strip().split()[0]
            newname = 'word_%d.jpg' % ((num // 20) + 1)
            img_dir = syn_dir + 'splitimg/%d/' % ((num % 20) + 1)
            gt_dir = syn_dir + 'splitimg/gt/'
            gt_name = 'gt_%d.txt' % ((num % 20) + 1)
            if not os.path.exists(img_dir):
                os.makedirs(img_dir)
            if not os.path.exists(gt_dir):
                os.makedirs(gt_dir)
            try:
                shutil.copy(syn_dir+imgname[2:], img_dir+newname)
            except:
                print('copy image failed!')
                continue
            label = imgname.split('_')[1]
            newline = newname+','+label+'\n'
            with open(gt_dir+gt_name, 'a') as f:
                f.write(newline)
            num += 1
            print(newline)


if __name__ == '__main__':
    fire.Fire()