-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathreader.py
316 lines (275 loc) · 13.3 KB
/
reader.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
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
import re
from random import randint
def read_case_setup(launch_filepath):
file = open(launch_filepath, 'r')
data = file.read()
data = re.sub('%.*\n','',data)
class setup:
pass
casedata = setup()
casedata.case_dir = None
casedata.analysis = dict.fromkeys(['type', 'import'], None)
casedata.training_parameters = {}
casedata.img_resize = [None,None]
casedata.img_processing = {'rotation': [None, None, None, None],
'translation': [None, None, None],
'zoom': [None, None],
'filter': [None, None, None],
'flip': [None, None]
}
casedata.data_augmentation = [None, None]
casedata.activation_plotting = {'n_samples': None, 'n_cols': None, 'rows2cols_ratio': None}
casedata.prediction = {'n_samples': None}
############################################### Data directory #####################################################
match = re.search('DATADIR\s*=\s*(.*).*', data)
if match:
casedata.case_dir = match.group(1)
################################################## Analysis ########################################################
casedata.analysis['case_ID'] = randint(1,9999)
# Type of analysis
match = re.search('TYPEANALYSIS\s*=\s*(\w+).*', data)
if match:
casedata.analysis['type'] = str.lower(match.group(1))
# Import
match = re.search('IMPORTMODEL\s*=\s*(\d).*', data)
if match:
casedata.analysis['import'] = int(match.group(1))
## Dataset augmentation
match = re.search('AUGDATA\s*=\s*(\d).*', data)
if match:
casedata.data_augmentation[0] = int(match.group(1))
match_factor = re.search('AUGDATASIZE\s*=\s*(\d+\.?\d*).*', data)
if match_factor:
casedata.data_augmentation[1] = float(match_factor.group(1))
############################################# Training parameters ##################################################
# Training dataset size
match = re.search('TRAINSIZE\s*=\s*(\d+\.?\d*|NONE).*', data)
if match:
if match.group(1) == 'NONE':
casedata.training_parameters['train_size'] = 0.75
else:
casedata.training_parameters['train_size'] = float(match.group(1))
# Learning rate
match = re.search('LEARNINGRATE\s*=\s*(.*)', data)
if match:
matches = re.findall('(\d+\.?\d*)', match.group(1))
casedata.training_parameters['learning_rate'] = float(matches[0]) if len(matches) == 1 else [float(item) for
item in matches]
# L2 regularizer
match = re.search('L2REG\s*=\s*(.*|NONE)', data)
if match:
matches = re.findall('(\d+\.?\d*)', match.group(1))
if matches:
casedata.training_parameters['l2_reg'] = float(matches[0]) if len(matches) == 1 else [float(item) for item
in matches]
else:
casedata.training_parameters['l2_reg'] = 0.0
# L1 regularizer
match = re.search('L1REG\s*=\s*(.*|NONE)', data)
if match:
matches = re.findall('(\d+\.?\d*)', match.group(1))
if matches:
casedata.training_parameters['l1_reg'] = float(matches[0]) if len(matches) == 1 else [float(item) for item
in matches]
else:
casedata.training_parameters['l1_reg'] = 0.0
# Dropout
match = re.search('DROPOUT\s*=\s*(.*|NONE)', data)
if match:
matches = re.findall('(\d+\.?\d*)', match.group(1))
if matches:
casedata.training_parameters['dropout'] = float(matches[0]) if len(matches) == 1 else [float(item) for item
in matches]
else:
casedata.training_parameters['dropout'] = 0.0
# Number of epochs
match = re.search('EPOCHS\s*=\s*(\d+|NONE).*', data)
if match:
if match.group(1) == 'NONE':
casedata.training_parameters['epochs'] = 1
else:
casedata.training_parameters['epochs'] = int(match.group(1))
# Batch size
match = re.search('BATCHSIZE\s*=\s*(\d+\.?\d*|NONE).*', data)
if match:
if match.group(1) == 'NONE':
casedata.training_parameters['batch_size'] = None
else:
casedata.training_parameters['batch_size'] = int(match.group(1))
# Activation function
match = re.search('ACTIVATION\s*=\s*((\w+)\s*,?\s*(\w+)*)\s*.*', data)
if match:
if None in match.groups():
casedata.training_parameters['activation'] = str.lower(match.group(2))
else:
casedata.training_parameters['activation'] = [str.lower(item) for item in match.groups()[1:]]
######################################## Image processing parameters ###############################################
# Image resize
match_dist = re.search('IMAGERESIZE\s*=\s*\((\d+|NONE)\,+(\d+|NONE)\).*', data)
if match_dist:
casedata.img_resize[0] = int(match_dist.group(1))
casedata.img_resize[1] = int(match_dist.group(2))
casedata.img_resize = tuple(casedata.img_resize)
# Rotation
match = re.search('ROTATION\s*=\s*(\d).*', data)
if match:
casedata.img_processing['rotation'][0] = int(match.group(1))
match_angle = re.search('ROTATIONANGLE\s*=\s*([\+|\-]?\d+\.?\d*).*', data)
if match_angle:
casedata.img_processing['rotation'][1] = float(match_angle.group(1))
match = re.search('ROTATIONCENTER\s*=\s*\((\d+|NONE)\,+(\d+|NONE)\).*', data)
if match:
if match.group(1) != 'NONE':
casedata.img_processing['rotation'][2] = int(match.group(1))
elif match.group(2) != 'NONE':
casedata.img_processing['rotation'][3] = int(match.group(2))
# Translation
match = re.search('TRANSLATION\s*=\s*(\d).*', data)
if match:
casedata.img_processing['translation'][0] = int(match.group(1))
match_dist = re.search('TRANSLATIONDIST\s*=\s*\(([\+|\-]?\d+|NONE)\,+([\+|\-]?\d+|NONE)\).*', data)
if match_dist:
casedata.img_processing['translation'][1] = float(match_dist.group(1))
casedata.img_processing['translation'][2] = float(match_dist.group(2))
# Zoom
match = re.search('ZOOM\s*=\s*(\d).*', data)
if match:
casedata.img_processing['zoom'][0] = int(match.group(1))
match_factor = re.search('ZOOMFACTOR\s*=\s*(\d+\.?\d*).*', data)
if match_factor:
casedata.img_processing['zoom'][1] = float(match_factor.group(1))
# Filter
match = re.search('FILTER\s*=\s*(\d).*', data)
if match:
casedata.img_processing['filter'][0] = int(match.group(1))
match_type = re.search('FILTERTYPE\s*=\s*(\w+).*', data)
casedata.img_processing['filter'][1] = str.lower(match_type.group(1))
if match_type:
filtertype = str.lower(match_type.group(1))
if filtertype == 'gaussian' or filtertype == 'median':
filter_param = re.search(
'FILTERPARAM\s*=\s*\(\s*SIZE\s*\,\s*(\d+|NONE)\s*\,\s*SIGMA\s*\,\s*(\d+|NONE)\s*\).*', data)
casedata.img_processing['filter'][2] = {}
casedata.img_processing['filter'][2]['ksize'] = int(filter_param.group(1))
casedata.img_processing['filter'][2]['sigma'] = int(filter_param.group(2)) if filtertype == 'gaussian' else None
elif str.lower(match_type.group(1)) == 'bilateral':
filter_param = re.search(
'FILTERPARAM\s*=\s*\(\s*D\s*\,\s*(\d+|NONE)\s*\,\s*SIGMACOLOR\s*\,\s*(\d+|NONE)\s*\,\s*SIGMASPACE\s*\,\s*(\d+|NONE)\s*\).*',
data)
casedata.img_processing['filter'][2] = {}
casedata.img_processing['filter'][2]['d'] = int(filter_param.group(1))
casedata.img_processing['filter'][2]['sigmaColor'] = int(filter_param.group(2))
casedata.img_processing['filter'][2]['sigmaSpace'] = int(filter_param.group(3))
# Flip
match = re.search('FLIP\s*=\s*(\d).*', data)
if match:
casedata.img_processing['flip'][0] = int(match.group(1))
match_type = re.search('FLIPTYPE\s*=\s*(\w+).*', data)
if match_type:
casedata.img_processing['flip'][1] = str.lower(match_type.group(1))
######################################### Activation plotting parameters ###########################################
# Number of samples
match = re.search('NSAMPLESACT\s*=\s*(\d+).*', data)
if match:
casedata.activation_plotting['n_samples'] = int(match.group(1))
# Number of columns
match = re.search('NCOLS\s*=\s*(\d+).*', data)
if match:
casedata.activation_plotting['n_cols'] = int(match.group(1))
# Rows-to-columns figure ratio
match = re.search('ROWS2COLS\s*=\s*(\d+).*', data)
if match:
casedata.activation_plotting['rows2cols_ratio'] = int(match.group(1))
############################################ Prediction parameters #################################################
match = re.search('DATAPREDDIR\s*=\s*(.*).*',data)
if match:
casedata.prediction['dir'] = match.group(1)
return casedata
def read_case_logfile(log_filepath):
file = open(log_filepath, 'r')
data = file.read()
data = re.sub('%.*\n','', data)
class setup:
pass
casedata = setup()
casedata.analysis = dict.fromkeys(['case_ID','type', 'import'], None)
casedata.training_parameters = dict()
casedata.img_resize = [None,None]
################################################## Analysis ########################################################
# Case ID
match = re.search('CASE ID\s*=\s*(\d+\.?\d*|NONE).*', data)
if match:
casedata.analysis['case_ID'] = int(match.group(1))
# Type of analysis
match = re.search('ANALYSIS\s*=\s*(\w+).*', data)
if match:
casedata.analysis['type'] = str.lower(match.group(1))
# Image shape
match = re.search('INPUT SHAPE\s*=\s*\((.*)\).*', data)
if match:
casedata.img_size = [int(item) for item in re.findall('\d+',match.group(1))]
# Import
match = re.search('IMPORTED MODEL\s*=\s*(\d).*', data)
if match:
casedata.analysis['import'] = int(match.group(1))
############################################# Training parameters ##################################################
# Training dataset size
match = re.search('TRAINING SIZE\s*=\s*(\d+\.?\d*|NONE).*', data)
if match:
if match.group(1) == 'NONE':
casedata.training_parameters['train_size'] = 0.75
else:
casedata.training_parameters['train_size'] = float(match.group(1))
# Learning rate
match = re.search('LEARNING RATE\s*=\s*\[*(.*)\]*', data)
if match:
matches = re.findall('(\d+\.?\d*)',match.group(1))
casedata.training_parameters['learning_rate'] = float(matches[0]) if len(matches) == 1 else [float(item) for item in matches]
# L2 regularizer
match = re.search('L2 REGULARIZER\s*=\s*\[*(.*|NONE)\]*', data)
if match:
matches = re.findall('(\d+\.?\d*)',match.group(1))
if matches:
casedata.training_parameters['l2_reg'] = float(matches[0]) if len(matches) == 1 else [float(item) for item in matches]
else:
casedata.training_parameters['l2_reg'] = 0.0
# L1 regularizer
match = re.search('L1 REGULARIZER\s*=\s*\[*(.*|NONE)\]*', data)
if match:
matches = re.findall('(\d+\.?\d*)',match.group(1))
if matches:
casedata.training_parameters['l1_reg'] = float(matches[0]) if len(matches) == 1 else [float(item) for item in matches]
else:
casedata.training_parameters['l1_reg'] = 0.0
# Dropout
match = re.search('DROPOUT\s*=\s*\[*(.*|NONE)\]*', data)
if match:
matches = re.findall('(\d+\.?\d*)',match.group(1))
if matches:
casedata.training_parameters['dropout'] = float(matches[0]) if len(matches) == 1 else [float(item) for item in matches]
else:
casedata.training_parameters['dropout'] = 0.0
# Number of epochs
match = re.search('NUMBER OF EPOCHS\s*=\s*(\d+\.?\d*|NONE).*', data)
if match:
if match.group(1) == 'NONE':
casedata.training_parameters['epochs'] = 1
else:
casedata.training_parameters['epochs'] = int(match.group(1))
# Batch size
match = re.search('BATCH SIZE\s*=\s*(\d+\.?\d*|NONE).*', data)
if match:
if match.group(1) == 'NONE':
casedata.training_parameters['batch_size'] = None
else:
casedata.training_parameters['batch_size'] = int(match.group(1))
# Activation function
match = re.search('ACTIVATION\s*=\s*\[*(.*)\]*\s*.*', data)
if match:
matches = re.findall('(\w+)',match.group(1))
if matches:
if len(matches) == 1:
casedata.training_parameters['activation'] = str.lower(matches[0])
else:
casedata.training_parameters['activation'] = [str.lower(item) for item in matches]
return casedata