-
Notifications
You must be signed in to change notification settings - Fork 6
/
utils.py
513 lines (410 loc) · 17.5 KB
/
utils.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
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
import os
import sys
import pickle
import numpy as np
import xml.etree.ElementTree as ET
import random
import svgwrite
from IPython.display import SVG, display
import tensorflow as tf
def get_bounds( data, factor ):
min_x = 0
max_x = 0
min_y = 0
max_y = 0
abs_x = 0
abs_y = 0
for i in range( len( data ) ):
x = float( data[ i, 0 ] )/factor
y = float( data[ i, 1 ] )/factor
abs_x += x
abs_y += y
min_x = min( min_x, abs_x )
min_y = min( min_y, abs_y )
max_x = max( max_x, abs_x )
max_y = max( max_y, abs_y )
return ( min_x, max_x, min_y, max_y )
# version where each path is entire stroke ( smaller svg size, but have to keep same color )
def draw_strokes( data, factor = 10, svg_filename = 'sample.svg' ):
min_x, max_x, min_y, max_y = get_bounds( data, factor )
dims = ( 50 + max_x - min_x, 50 + max_y - min_y )
dwg = svgwrite.Drawing( svg_filename, size = dims )
dwg.add( dwg.rect( insert = ( 0, 0 ), size = dims, fill = 'white' ) )
lift_pen = 1
abs_x = 25 - min_x
abs_y = 25 - min_y
p = "M%s, %s " % ( abs_x, abs_y )
command = "m"
for i in range( len( data ) ):
if ( lift_pen == 1 ):
command = "m"
elif ( command != "l" ):
command = "l"
else:
command = ""
x = float( data[ i, 0 ] )/factor
y = float( data[ i, 1 ] )/factor
lift_pen = data[ i, 2 ]
p += command+str( x )+", "+str( y )+" "
the_color = "black"
stroke_width = 1
dwg.add( dwg.path( p ).stroke( the_color, stroke_width ).fill( "none" ) )
dwg.save( )
display( SVG( dwg.tostring( ) ) )
def draw_strokes_eos_weighted( stroke, param, factor = 10, svg_filename = 'sample[ A_eos.svg' ):
c_data_eos = np.zeros( ( len( stroke ), 3 ) )
for i in range( len( param ) ):
c_data_eos[ i, : ] = ( 1-param[ i ][ 6 ][ 0 ] )*225 # make color gray scale, darker = more likely to eos
draw_strokes_custom_color( stroke, factor = factor, svg_filename = svg_filename, color_data = c_data_eos, stroke_width = 3 )
def draw_strokes_random_color( stroke, factor = 10, svg_filename = 'sample_random_color.svg', per_stroke_mode = True ):
c_data = np.array( np.random.rand( len( stroke ), 3 )*240, dtype = np.uint8 )
if per_stroke_mode:
switch_color = False
for i in range( len( stroke ) ):
if switch_color == False and i > 0:
c_data[ i ] = c_data[ i-1 ]
if stroke[ i, 2 ] < 1: # same strike
switch_color = False
else:
switch_color = True
draw_strokes_custom_color( stroke, factor = factor, svg_filename = svg_filename, color_data = c_data, stroke_width = 2 )
def draw_strokes_custom_color( data, factor = 10, svg_filename = 'test.svg', color_data = None, stroke_width = 1 ):
min_x, max_x, min_y, max_y = get_bounds( data, factor )
dims = ( 50 + max_x - min_x, 50 + max_y - min_y )
dwg = svgwrite.Drawing( svg_filename, size = dims )
dwg.add( dwg.rect( insert = ( 0, 0 ), size = dims, fill = 'white' ) )
lift_pen = 1
abs_x = 25 - min_x
abs_y = 25 - min_y
for i in range( len( data ) ):
x = float( data[ i, 0 ] )/factor
y = float( data[ i, 1 ] )/factor
prev_x = abs_x
prev_y = abs_y
abs_x += x
abs_y += y
if ( lift_pen == 1 ):
p = "M "+str( abs_x )+", "+str( abs_y )+" "
else:
p = "M +"+str( prev_x )+", "+str( prev_y )+" L "+str( abs_x )+", "+str( abs_y )+" "
lift_pen = data[ i, 2 ]
the_color = "black"
if ( color_data is not None ):
the_color = "rgb( "+str( int( color_data[ i, 0 ] ) )+", "+str( int( color_data[ i, 1 ] ) )+", "+str( int( color_data[ i, 2 ] ) )+" )"
dwg.add( dwg.path( p ).stroke( the_color, stroke_width ).fill( the_color ) )
dwg.save( )
display( SVG( dwg.tostring( ) ) )
def draw_strokes_pdf( data, param, factor = 10, svg_filename = 'sample_pdf.svg' ):
min_x, max_x, min_y, max_y = get_bounds( data, factor )
dims = ( 50 + max_x - min_x, 50 + max_y - min_y )
dwg = svgwrite.Drawing( svg_filename, size = dims )
dwg.add( dwg.rect( insert = ( 0, 0 ), size = dims, fill = 'white' ) )
abs_x = 25 - min_x
abs_y = 25 - min_y
num_mixture = len( param[ 0 ][ 0 ] )
for i in range( len( data ) ):
x = float( data[ i, 0 ] )/factor
y = float( data[ i, 1 ] )/factor
for k in range( num_mixture ):
pi = param[ i ][ 0 ][ k ]
if pi > 0.01: # optimisation, ignore pi's less than 1% chance
mu1 = param[ i ][ 1 ][ k ]
mu2 = param[ i ][ 2 ][ k ]
s1 = param[ i ][ 3 ][ k ]
s2 = param[ i ][ 4 ][ k ]
sigma = np.sqrt( s1*s2 )
dwg.add( dwg.circle( center = ( abs_x+mu1*factor, abs_y+mu2*factor ), r = int( sigma*factor ) ).fill( 'red', opacity = pi/( sigma*sigma*factor ) ) )
prev_x = abs_x
prev_y = abs_y
abs_x += x
abs_y += y
dwg.save( )
display( SVG( dwg.tostring( ) ) )
class DataLoader( ):
def getRandValue( self ):
value = self.randvalues[ self.randvaluepointer ]
self.randvaluepointer += 1
if ( self.randvaluepointer >= self.nrrandvalues ):
self.randvaluepointer = 0
return value
def createRandValues( self ):
self.nrrandvalues = 1000
self.randvalues = np.zeros( ( self.nrrandvalues ), dtype = np.float32 )
for i in range( self.nrrandvalues ):
value = random.random( )
self.randvalues[ i ] = value
self.randvaluepointer = 0
def getClassLabels( self ):
if self.train:
fn = self.data_dir + "trainlabels.txt"
else:
fn = self.data_dir + "testlabels.txt"
classlabels = np.loadtxt( fn )
classlabels = classlabels[ :self.nrinputfiles ]
return classlabels
def findAvailableExamples( self, args ):
self.availableExamples = [ ]
findexamples = True
if findexamples:
for i in range( len( self.classlabels ) ):
if ( self.classlabels[ i ] < args.curnrdigits ):
self.availableExamples.append( i )
self.availableExamples = np.array( self.availableExamples )
def __init__( self, datadir, args, totnrfiles, curnrexamples, seqlength = 0, train = 1, file_label = "", print_input = 0, rangemin = 0, rangelen = 0 ):
random.seed( 100*args.runnr )
np.random.seed( 100*args.runnr )
tf.set_random_seed( 100*args.runnr )
self.args = args
self.data_dir = datadir
self.train = train
if self.train:
self.traintest = "train"
else:
self.traintest = "test"
self.rangemin = rangemin
self.rangelen = rangelen
self.nrinputfiles = totnrfiles
self.curnrexamples = curnrexamples
self.nrseq_per_batch = args.nrseq_per_batch
self.file_label = file_label
self.print_input = print_input
self.nrinputvars_data = self.getInputVectorLength( args )
self.max_seq_length = args.max_seq_length
self.nrsequenceinputs = 4 #dx dy eos eod
self.nrauxinputvars = args.nrClassOutputVars #either [ 0..9 dx dy eos eod ] or [ dx dy eos ]
strokedatafile = os.path.join( self.data_dir, "strokes_"+self.traintest+"ing_data"+ file_label+args.explabel+ ".cpkl" )
raw_data_dir = self.data_dir+"/lineStrokes"
print ( "creating data cpkl file from source data" )
self.preprocess( args, raw_data_dir, strokedatafile )
if ( seqlength > 0 ): #provided
self.seq_length = seqlength
else:
self.seq_length = min( self.max_seq_length, args.maxdigitlength_nrpoints )
self.load_preprocessed( args, strokedatafile )
self.classlabels = self.getClassLabels( )
self.findAvailableExamples( args )
self.nrbatches_per_epoch = max( 1, int( self.curnrexamples / self.nrseq_per_batch ) )
print ( "curnrexamples", self.curnrexamples, "seq_length", self.seq_length, " --> nrbatches_per_epoch: ", self.nrbatches_per_epoch )
print ( "loaded data" )
self.reset_batch_pointer( args )
def constructInputFileName( self, args, file_label, imgnr ):
filename = self.data_dir + self.traintest + 'img' + file_label + '-' + str( imgnr ) + '-targetdata.txt' #currently, we expect 14 inputs
return filename
def getInputVectorLength( self, args ):
result = [ ]
filename = self.constructInputFileName( args, self.file_label, imgnr = 0 )
with open( filename ) as f:
points = [ ]
line = f.readline( )
print ( "read sample line from inputdata file: ", line )
nrs = [ float( x ) for x in line.split( ) ]
length = len( nrs )
print ( "Determined nrinputvars based on data: ", length )
self.nrinputvars_data = length
return length
def preprocess( self, args, data_dir, strokedatafile ):
filelist = [ ]
if len( args.fileselection )>0:
fileselection = ' '.join( args.fileselection )
if len( fileselection )>0:
fileselection = [ int( s ) for s in fileselection.split( ', ' ) ]
for imgnr in range( 0, self.nrinputfiles ):
if len( args.fileselection )>0:
fname = self.constructInputFileName( args, self.file_label, fileselection[ imgnr ] )
else:
fname = self.constructInputFileName( args, self.file_label, imgnr )
filelist.append( fname )
def getStrokes( filename, nrauxinputvars ): #returns array of arrays with points
result_points = [ ]
result_auxinputs = [ ]
nrsequencevars = 4
dxmin = 1e100
dxmax = -1e100
dymin = 1e100
dymax = -1e100
nrauxinputs_data = 10
with open( filename ) as f:
points = [ ]
auxinputs = [ ]
for line in f: # read rest of lines
nrs = [ float( x ) for x in line.split( ) ]
auxinputvalues = nrs[ 0:nrauxinputvars ]
point = nrs[ nrauxinputs_data:nrauxinputs_data+nrsequencevars ] #currently: x, y, end-of-stroke
points.append( point )
auxinputs.append( auxinputvalues )
result_points.append( points )
result_auxinputs.append( auxinputs )
pointarray = np.array( points )
digitlength_nrpoints = len( points )
dxmin = pointarray[ :, 0 ].min( )
dxmax = pointarray[ :, 0 ].max( )
dymin = pointarray[ :, 1 ].min( )
dymax = pointarray[ :, 1 ].max( )
ranges = [ dxmin, dxmax, dymin, dymax ]
return result_auxinputs, result_points, ranges, digitlength_nrpoints
# converts a list of arrays into a 2d numpy int16 array
def convert_stroke_to_array( stroke ):
n_point = 0
for i in range( len( stroke ) ):
n_point += len( stroke[ i ] )
prev_x = 0
prev_y = 0
counter = 0
nrsequencevars = 4
stroke_data = np.zeros( ( n_point, nrsequencevars ), dtype = np.int16 )
for j in range( len( stroke ) ):
for k in range( len( stroke[ j ] ) ):
for s in range( nrsequencevars ):
stroke_data[ counter, s ] = int( stroke[ j ][ k ][ s ] )
counter += 1
return stroke_data
# converts a list of arrays into a 2d numpy int16 array
def convert_auxinputs_to_array( auxinputs, nrauxinputvars ):
n_point = 0
for i in range( len( auxinputs ) ):
n_point += len( auxinputs[ i ] )
auxinputdata = np.zeros( ( n_point, nrauxinputvars ), dtype = np.int16 )
prev_x = 0
prev_y = 0
counter = 0
for j in range( len( auxinputs ) ):
for k in range( len( auxinputs[ j ] ) ):
for a in range( nrauxinputvars ):
auxinputdata[ counter, a ] = int( auxinputs[ j ][ k ][ a ] )
counter += 1
return auxinputdata
# preprocess body: build stroke array
strokearray = [ ]
auxinputarray = [ ]
rangelist = [ ]
self.seqlengthlist = [ ]
if self.train:
args.maxdigitlength_nrpoints = 0
digitlengthsum = 0
for i in range( len( filelist ) ):
print ( 'dataloader', self.traintest, 'processing '+filelist[ i ] )
[ auxinputs, strokeinputs, ranges, digitlength_nrpoints ] = getStrokes( filelist[ i ], self.nrauxinputvars )
strokearray.append( convert_stroke_to_array( strokeinputs ) )
auxinputarray.append( convert_auxinputs_to_array( auxinputs, self.nrauxinputvars ) )
rangelist.append( ranges )
self.seqlengthlist.append( digitlength_nrpoints )
if self.train:
args.maxdigitlength_nrpoints = max( args.maxdigitlength_nrpoints, digitlength_nrpoints )
digitlengthsum += digitlength_nrpoints
rangearray = np.array( rangelist )
ranges = [ rangearray[ :, 0 ].min( ), rangearray[ :, 1 ].max( ), rangearray[ :, 2 ].min( ), rangearray[ :, 3 ].max( ) ]
print ( "found overall ranges", ranges )
self.avgseqlength = digitlengthsum / len( filelist )
print( "dataloader: found avg seq length: ", self.avgseqlength )
print ( "found maxdigitlength_nrpoints", args.maxdigitlength_nrpoints )
f = open( strokedatafile, "wb" )
pickle.dump( strokearray, f )
pickle.dump( auxinputarray, f )
pickle.dump( ranges, f )
pickle.dump( self.seqlengthlist, f )
f.close( )
def load_preprocessed( self, args, strokedatafile ):
f = open( strokedatafile, "rb" )
self.strokedataraw = pickle.load( f )
self.auxdataraw = pickle.load( f )
self.ranges = pickle.load( f )
self.seqlengthlist = pickle.load( f )
f.close( )
print ( "loaded ranges", self.ranges )
print ( "rangemin", self.rangemin, "rangelen", self.rangelen )
self.strokedata = [ ] #contains one array per file
self.auxdata = [ ]
counter = 0
for data_el in self.strokedataraw:
data = np.array( np.zeros( ( self.seq_length, self.nrsequenceinputs ), dtype = np.float32 ) )
len_data = len( data )
nrpoints = min( self.seq_length, len( data_el ) )
data[ :nrpoints, ] = data_el[ :nrpoints ]
if ( len( data_el ) > self.seq_length ) and ( self.seq_length >= args.max_seq_length ):
data[ self.seq_length-1, 2:4 ] = np.ones( ( 1, 2 ), dtype = np.float32 ) #add eos and eod for sequences exceeding length
data[ nrpoints:, 0:4 ] = np.zeros( ( len_data - nrpoints, 4 ), dtype = np.float32 ) #pad remainder with zero rows
data[ :, 0:2 ] -= self.rangemin
data[ :, 0:2 ] /= self.rangelen
self.strokedata.append( data )
counter += 1
for data_el in self.auxdataraw:
data = np.array( np.zeros( ( self.seq_length, self.nrauxinputvars ), dtype = np.float32 ) )
nrpoints = min( self.seq_length, len( data_el ) )
data[ :nrpoints, ] = data_el[ :nrpoints ]
data[ nrpoints:self.seq_length, ] = data[ nrpoints-1, ]
self.auxdata.append( data )
print ( "#sequences found in data: ", counter )
def next_batch( self, args, curseqlength ):
# returns a batch of the training data of nrseq_per_batch * seq_length points
x_batch = [ ]
y_batch = [ ]
seqlen = self.seq_length
sequence_index = [ ]
use_points_stopcrit = False
nrpoints_per_batch = 0
if hasattr( args, 'nrpoints_per_batch' ):
nrpoints_per_batch = args.nrpoints_per_batch
if nrpoints_per_batch > 0:
use_points_stopcrit = True
batch_nrpoints = 0
batch_sequencenr = 0
done = False
while not done:
sequence_index.append( self.pointer )
strokes = np.copy( self.strokedata[ self.pointer ] )
auxvalues = np.copy( self.auxdata[ self.pointer ] )
if args.useStrokeOutputVars:
ytab = np.copy( np.hstack( [ auxvalues[ 1:seqlen ], strokes[ 1:seqlen ] ] ) )
else:
ytab = np.copy( np.hstack( [ auxvalues[ 1:seqlen ] ] ) )
if args.discard_classvar_inputs:
auxvalues[ : ] = 0
if args.useClassInputVars:
xtab = np.hstack( [ auxvalues[ :seqlen-1 ], strokes[ :seqlen-1 ] ] )
else:
xtab = strokes[ :seqlen-1 ]
actual_seq_length = self.seqlengthlist[ self.pointer ]
firsttrainstep = 0
if hasattr( args, 'firsttrainstep' ):
firsttrainstep = args.firsttrainstep
firsttrainstep = min ( firsttrainstep, actual_seq_length - 1 )
if firsttrainstep > 0: #remove earlier part from _target_ data, so that it will not be used in loss.
ytab[ :firsttrainstep, : ] = 0
#only keep points up to current seq_length - 1; e.g. if sequence has 3 points, use 2 pairs of ( x, y ): k = 1.n-1 for x and k = 2..n for y
firstafter = min( actual_seq_length - 1, curseqlength ) #zero out part after sequence
xtab[ firstafter:, : ] = 0
ytab[ firstafter:, : ] = 0
nrusedpoints = firstafter - firsttrainstep
if args.discard_inputs:
xtab[ : ] = 0
x_batch.append( np.copy( xtab ) )
y_batch.append( np.copy( ytab ) )
self.next_batch_pointer( args )
batch_sequencenr += 1
nrseq_per_batch = self.nrseq_per_batch
if ( not self.train ) and hasattr( args, 'nrseq_per_batch_test' ):
nrseq_per_batch = args.nrseq_per_batch_test
if use_points_stopcrit:
batch_nrpoints += nrusedpoints
done = batch_nrpoints >= nrpoints_per_batch
else:
done = batch_sequencenr >= nrseq_per_batch
return x_batch, y_batch, sequence_index
def selectExamples( self, nrdigits ):
sample = np.random.permutation( len( self.availableExamples ) )
return self.availableExamples[ sample ]
def next_batch_pointer( self, args ):
self.index += 1
if ( self.index >= len( self.example_permutation ) ):
self.reset_batch_pointer( args )
self.pointer = self.example_permutation[ self.index ]
def reset_batch_pointer( self, args ):
self.index = 0
if ( args.incremental_nr_digits and self.train ):
self.example_permutation = self.selectExamples( args.curnrdigits )
else:
if self.train:
self.example_permutation = np.random.permutation( int( self.curnrexamples ) )
else:
self.example_permutation = np.arange( 0, int( self.curnrexamples ) )
self.pointer = self.example_permutation[ self.index ]