forked from knaw-huc/loghi
-
Notifications
You must be signed in to change notification settings - Fork 0
/
modified-inference-pipeline.sh
468 lines (385 loc) · 15.1 KB
/
modified-inference-pipeline.sh
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
#!/bin/bash
VERSION=2.1.2
set -e
STOPONERROR=1
## TOGGLES: set 1 = on, 0 = off to add/ remove recognition steps from the pipeline
# Baseline Recognition:
BASELINELAYPA=1
# Region (Field) Recognition (only pick one of these two)
# AI-powered (trained) Region Recognition:
REGIONLAYPA=0
# Rule-based Region Recognition (needs Baselines to work):
REGIONRULEBASED=1
# Text Recognition
HTRLOGHI=0
# Recalculate Reading Order (custom region reading order recalculation script)
NEWREADINGORDER=1
## MODEL SELECTION
# specify model paths (only needed if corresponding recognition step is set to on)
# Baseline
LAYPABASELINEMODEL=INSERT_FULL_PATH_TO_YAML_HERE
LAYPABASELINEMODELWEIGHTS=INSERT_FULLPATH_TO_PTH_HERE
# Region
LAYPAREGIONMODEL=INSERT_FULL_PATH_TO_YAML_HERE
LAYPAREGIONMODELWEIGHTS=INSERT_FULLPATH_TO_PTH_HERE
# Text Recognition
HTRLOGHIMODEL=INSERT_FULL_PATH_TO_LOGHI_HTR_MODEL_HERE
# Further settings for REGIONRULEBASED
# If the edge of baseline is closer than x pixels...
REGIONRULEBASEDBORDERMARGIN=0
# Clean borders if 1
REGIONRULEBASEDCLEANBORDERS=0
# How many threads to use for recalculating reading order
REGIONRULEBASEDTHREADS=4
# Detect language of Text Lines
DETECTLANGUAGE=1
# Interpolate word locations
SPLITWORDS=1
# BEAMWIDTH: higher makes results slightly better at the expense of lot of computation time. In general don't set higher than 10
BEAMWIDTH=4
# Used gpu ids, set to "-1" to use CPU, "0" for first, "1" for second, etc
GPU=0
# Use 2013 PageXML namespace
USE2013NAMESPACE=1
# DO NOT MODIFY BELOW THIS LINE
# ------------------------------
function check_error_and_exit {
# $1 - The error message to display
# $2 - The status of the last command executed before the function was called
if [[ $STOPONERROR -eq 1 && $2 -ne 0 ]]; then
echo "$1 has failed"
exit 1
fi
}
# Check for proper command-line arguments
if [ "$#" -ne 1 ]; then
echo "Usage: $0 <path_to_images>"
exit 1
fi
tmpdir=$(mktemp -d)
echo "Temporary directory created at: $tmpdir"
mkdir -p $tmpdir/imagesnippets/
mkdir -p $tmpdir/output
DOCKERLOGHITOOLING=loghi/docker.loghi-tooling:$VERSION
DOCKERLAYPA=loghi/docker.laypa:$VERSION
DOCKERLOGHIHTR=loghi/docker.htr:$VERSION
DOCKERGPUPARAMS=""
if [[ $GPU -gt -1 ]]; then
DOCKERGPUPARAMS="--gpus device=${GPU}"
echo "Using GPU ${GPU}"
fi
IMAGES_PATH=`realpath $1`
# Housekeeping: remove any existing *.done files
find $IMAGES_PATH -name '*.done' -exec rm -f "{}" \;
if [[ $USE2013NAMESPACE -eq 1 ]]; then
namespace=" -use_2013_namespace "
fi
# First step: Laypa
if [[ $BASELINELAYPA -eq 1 ]]; then
echo "Running Laypa baseline detection"
LAYPA_IN=$IMAGES_PATH
LAYPA_OUT=$IMAGES_PATH
LAYPADIR="$(dirname "${LAYPABASELINEMODEL}")"
docker run $DOCKERGPUPARAMS --rm -it -u $(id -u ${USER}):$(id -g ${USER}) -m 32000m --shm-size 10240m \
-v $LAYPADIR:$LAYPADIR \
-v $LAYPA_IN:$LAYPA_IN \
-v $LAYPA_OUT:$LAYPA_OUT \
$DOCKERLAYPA \
python run.py \
-c $LAYPABASELINEMODEL \
-i $LAYPA_IN \
-o $LAYPA_OUT \
--opts MODEL.WEIGHTS "" TEST.WEIGHTS $LAYPABASELINEMODELWEIGHTS | tee -a $tmpdir/log.txt
# Check if failed
status=$?
check_error_and_exit "Laypa baseline detection" $status
# Set as_single_region flag by default
as_single_region=" -as_single_region "
echo "Laypa baseline detection done"
if [[ $REGIONLAYPA -eq 1 ]]; then
echo "Running Laypa region detection"
docker run $DOCKERGPUPARAMS --rm -it -u $(id -u ${USER}):$(id -g ${USER}) -m 32000m --shm-size 10240m \
-v $LAYPADIR:$LAYPADIR \
-v $LAYPA_IN:$LAYPA_IN \
-v $LAYPA_OUT:$LAYPA_OUT \
$DOCKERLAYPA \
python run.py \
-c $LAYPAREGIONMODEL \
-i $LAYPA_IN \
-o $LAYPA_OUT \
--opts MODEL.WEIGHTS "" TEST.WEIGHTS $LAYPAREGIONMODELWEIGHTS | tee -a $tmpdir/log.txt
# Check if failed
status=$?
check_error_and_exit "Laypa region detection" $status
as_single_region=""
echo "Laypa region detection done"
fi
# Second step: extract baselines and regions
echo "Extracting baselines and regions"
docker run --rm -u $(id -u ${USER}):$(id -g ${USER}) \
-v $LAYPA_IN:$LAYPA_IN \
-v $LAYPA_OUT:$LAYPA_OUT \
$DOCKERLOGHITOOLING \
/src/loghi-tooling/minions/target/appassembler/bin/MinionExtractBaselines \
-input_path_image $LAYPA_IN \
-input_path_png $LAYPA_OUT/page/ \
-input_path_page $LAYPA_OUT/page/ \
-output_path_page $LAYPA_OUT/page/ \
-recalculate_textline_contours_from_baselines \
$as_single_region \
$namespace | tee -a $tmpdir/log.txt
# Check if failed
status=$?
check_error_and_exit "MinionExtractBaselines" $status
echo "Extracting baselines and regions done"
fi
# Third step: cut out snippets
if [[ $HTRLOGHI -eq 1 ]]; then
echo "Cutting out snippets"
docker run -u $(id -u ${USER}):$(id -g ${USER}) --rm \
-v $IMAGES_PATH/:$IMAGES_PATH/ \
-v $tmpdir:$tmpdir \
$DOCKERLOGHITOOLING /src/loghi-tooling/minions/target/appassembler/bin/MinionCutFromImageBasedOnPageXMLNew \
-input_path $IMAGES_PATH \
-outputbase $tmpdir/imagesnippets/ \
-output_type png \
-channels 4 \
-threads 4 $namespace| tee -a $tmpdir/log.txt
# Check if failed
status=$?
check_error_and_exit "MinionCutFromImageBasedOnPageXMLNew" $status
# Collect all the snippets in a file
find $tmpdir/imagesnippets/ -type f -name '*.png' > $tmpdir/lines.txt
echo "Running HTR"
LOGHIDIR="$(dirname "${HTRLOGHIMODEL}")"
echo docker run $DOCKERGPUPARAMS -u $(id -u ${USER}):$(id -g ${USER}) --rm -m 32000m --shm-size 10240m -ti \
-v /tmp:/tmp \
-v $tmpdir:$tmpdir \
-v $LOGHIDIR:$LOGHIDIR \
$DOCKERLOGHIHTR \
bash -c "LD_PRELOAD=/usr/lib/x86_64-linux-gnu/libtcmalloc_minimal.so.4 python3 /src/loghi-htr/src/main.py \
--model $HTRLOGHIMODEL \
--batch_size 64 \
--use_mask \
--inference_list $tmpdir/lines.txt \
--results_file $tmpdir/results.txt \
--gpu $GPU \
--output $tmpdir/output/ \
--beam_width $BEAMWIDTH " | tee -a $tmpdir/log.txt
docker run $DOCKERGPUPARAMS -u $(id -u ${USER}):$(id -g ${USER}) --rm -m 32000m --shm-size 10240m -ti \
-v /tmp:/tmp \
-v $tmpdir:$tmpdir \
-v $LOGHIDIR:$LOGHIDIR \
$DOCKERLOGHIHTR \
bash -c "LD_PRELOAD=/usr/lib/x86_64-linux-gnu/libtcmalloc_minimal.so.4 python3 /src/loghi-htr/src/main.py \
--model $HTRLOGHIMODEL \
--batch_size 64 \
--use_mask \
--inference_list $tmpdir/lines.txt \
--results_file $tmpdir/results.txt \
--gpu $GPU \
--output $tmpdir/output/ \
--beam_width $BEAMWIDTH " | tee -a $tmpdir/log.txt
# Check if failed
status=$?
check_error_and_exit "Loghi HTR" $status
echo "Loghi HTR done"
# Fourth step: merge results back into PageXML
docker run -u $(id -u ${USER}):$(id -g ${USER}) --rm \
-v $LOGHIDIR:$LOGHIDIR \
-v $IMAGES_PATH/:$IMAGES_PATH/ \
-v $tmpdir:$tmpdir \
$DOCKERLOGHITOOLING /src/loghi-tooling/minions/target/appassembler/bin/MinionLoghiHTRMergePageXML \
-input_path $IMAGES_PATH/page \
-results_file $tmpdir/results.txt \
-config_file $HTRLOGHIMODEL/config.json \
-htr_code_config_file $tmpdir/output/config.json \
$namespace | tee -a $tmpdir/log.txt
# Check if failed
status=$?
check_error_and_exit "MinionLoghiHTRMergePageXML" $status
echo "Merging HTR results into PageXML done"
fi
# Fifth step: Recalculate reading order
if [[ $REGIONRULEBASED -eq 1 ]]
then
echo "Recalculating reading order"
clean_borders=""
if [[ $REGIONRULEBASEDCLEANBORDERS -eq 1 ]]; then
echo "and cleaning borders"
clean_borders=" -clean_borders "
fi
docker run -u $(id -u ${USER}):$(id -g ${USER}) --rm \
-v $IMAGES_PATH/:$IMAGES_PATH/ \
-v $tmpdir:$tmpdir \
$DOCKERLOGHITOOLING /src/loghi-tooling/minions/target/appassembler/bin/MinionRecalculateReadingOrderNew \
-input_dir $IMAGES_PATH/page/ \
-border_margin $REGIONRULEBASEDBORDERMARGIN \
-threads $REGIONRULEBASEDTHREADS \
$clean_borders \
$namespace| tee -a $tmpdir/log.txt
# Check if failed
status=$?
check_error_and_exit "MinionRecalculateReadingOrder" $status
fi
if [[ $DETECTLANGUAGE -eq 1 ]]
then
echo "Detecting language..."
docker run -u $(id -u ${USER}):$(id -g ${USER}) --rm \
-v $IMAGES_PATH/:$IMAGES_PATH/ \
-v $tmpdir:$tmpdir \
$DOCKERLOGHITOOLING /src/loghi-tooling/minions/target/appassembler/bin/MinionDetectLanguageOfPageXml \
-page $IMAGES_PATH/page/ \
$namespace | tee -a $tmpdir/log.txt
# Check if failed
status=$?
check_error_and_exit "MinionDetectLanguageOfPageXml" $status
fi
if [[ $SPLITWORDS -eq 1 ]]
then
echo "MinionSplitPageXMLTextLineIntoWords..."
docker run -u $(id -u ${USER}):$(id -g ${USER}) --rm \
-v $IMAGES_PATH/:$IMAGES_PATH/ \
-v $tmpdir:$tmpdir \
$DOCKERLOGHITOOLING /src/loghi-tooling/minions/target/appassembler/bin/MinionSplitPageXMLTextLineIntoWords \
-input_path $IMAGES_PATH/page/ \
$namespace | tee -a $tmpdir/log.txt
# Check if failed
status=$?
check_error_and_exit "MinionSplitPageXMLTextLineIntoWords" $status
fi
if [[ $NEWREADINGORDER -eq 1 ]]
then
echo "Recalculating the Reading Order of the Regions..."
docker run $DOCKERGPUPARAMS -u $(id -u ${USER}):$(id -g ${USER}) --rm -it \
-v $IMAGES_PATH:$IMAGES_PATH/ \
-v $tmpdir:/tmp \
-v $(pwd):/workspace \
reorder-image \
bash -c "ls /workspace && python /workspace/reorder/reorder.py $IMAGES_PATH/page --overwrite"
# Check if failed
status=$?
check_error_and_exit "Recalculating the Reading Order of the regions" $status
echo "Finished Recalculating the Reading Order of the regions."
fi
# cleanup results
rm -rf $tmpdir
# Check if failed
status=$?
check_error_and_exit "MinionCutFromImageBasedOnPageXMLNew" $status
# Collect all the snippets in a file
find $tmpdir/imagesnippets/ -type f -name '*.png' > $tmpdir/lines.txt
echo "Running HTR"
LOGHIDIR="$(dirname "${HTRLOGHIMODEL}")"
echo docker run $DOCKERGPUPARAMS -u $(id -u ${USER}):$(id -g ${USER}) --rm -m 32000m --shm-size 10240m -ti \
-v /tmp:/tmp \
-v $tmpdir:$tmpdir \
-v $LOGHIDIR:$LOGHIDIR \
$DOCKERLOGHIHTR \
bash -c "LD_PRELOAD=/usr/lib/x86_64-linux-gnu/libtcmalloc_minimal.so.4 python3 /src/loghi-htr/src/main.py \
--model $HTRLOGHIMODEL \
--batch_size 64 \
--use_mask \
--inference_list $tmpdir/lines.txt \
--results_file $tmpdir/results.txt \
--gpu $GPU \
--output $tmpdir/output/ \
--beam_width $BEAMWIDTH " | tee -a $tmpdir/log.txt
docker run $DOCKERGPUPARAMS -u $(id -u ${USER}):$(id -g ${USER}) --rm -m 32000m --shm-size 10240m -ti \
-v /tmp:/tmp \
-v $tmpdir:$tmpdir \
-v $LOGHIDIR:$LOGHIDIR \
$DOCKERLOGHIHTR \
bash -c "LD_PRELOAD=/usr/lib/x86_64-linux-gnu/libtcmalloc_minimal.so.4 python3 /src/loghi-htr/src/main.py \
--model $HTRLOGHIMODEL \
--batch_size 64 \
--use_mask \
--inference_list $tmpdir/lines.txt \
--results_file $tmpdir/results.txt \
--gpu $GPU \
--output $tmpdir/output/ \
--beam_width $BEAMWIDTH " | tee -a $tmpdir/log.txt
# Check if failed
status=$?
check_error_and_exit "Loghi HTR" $status
echo "Loghi HTR done"
# Fourth step: merge results back into PageXML
docker run -u $(id -u ${USER}):$(id -g ${USER}) --rm \
-v $LOGHIDIR:$LOGHIDIR \
-v $IMAGES_PATH/:$IMAGES_PATH/ \
-v $tmpdir:$tmpdir \
$DOCKERLOGHITOOLING /src/loghi-tooling/minions/target/appassembler/bin/MinionLoghiHTRMergePageXML \
-input_path $IMAGES_PATH/page \
-results_file $tmpdir/results.txt \
-config_file $HTRLOGHIMODEL/config.json \
-htr_code_config_file $tmpdir/output/config.json \
$namespace | tee -a $tmpdir/log.txt
# Check if failed
status=$?
check_error_and_exit "MinionLoghiHTRMergePageXML" $status
echo "Merging HTR results into PageXML done"
fi
# Fifth step: Recalculate reading order
if [[ $REGIONRULEBASED -eq 1 ]]
then
echo "Recalculating reading order"
clean_borders=""
if [[ $REGIONRULEBASEDCLEANBORDERS -eq 1 ]]; then
echo "and cleaning borders"
clean_borders=" -clean_borders "
fi
docker run -u $(id -u ${USER}):$(id -g ${USER}) --rm \
-v $IMAGES_PATH/:$IMAGES_PATH/ \
-v $tmpdir:$tmpdir \
$DOCKERLOGHITOOLING /src/loghi-tooling/minions/target/appassembler/bin/MinionRecalculateReadingOrderNew \
-input_dir $IMAGES_PATH/page/ \
-border_margin $REGIONRULEBASEDBORDERMARGIN \
-threads $REGIONRULEBASEDTHREADS \
$clean_borders \
$namespace| tee -a $tmpdir/log.txt
# Check if failed
status=$?
check_error_and_exit "MinionRecalculateReadingOrder" $status
fi
if [[ $DETECTLANGUAGE -eq 1 ]]
then
echo "Detecting language..."
docker run -u $(id -u ${USER}):$(id -g ${USER}) --rm \
-v $IMAGES_PATH/:$IMAGES_PATH/ \
-v $tmpdir:$tmpdir \
$DOCKERLOGHITOOLING /src/loghi-tooling/minions/target/appassembler/bin/MinionDetectLanguageOfPageXml \
-page $IMAGES_PATH/page/ \
$namespace | tee -a $tmpdir/log.txt
# Check if failed
status=$?
check_error_and_exit "MinionDetectLanguageOfPageXml" $status
fi
if [[ $SPLITWORDS -eq 1 ]]
then
echo "MinionSplitPageXMLTextLineIntoWords..."
docker run -u $(id -u ${USER}):$(id -g ${USER}) --rm \
-v $IMAGES_PATH/:$IMAGES_PATH/ \
-v $tmpdir:$tmpdir \
$DOCKERLOGHITOOLING /src/loghi-tooling/minions/target/appassembler/bin/MinionSplitPageXMLTextLineIntoWords \
-input_path $IMAGES_PATH/page/ \
$namespace | tee -a $tmpdir/log.txt
# Check if failed
status=$?
check_error_and_exit "MinionSplitPageXMLTextLineIntoWords" $status
fi
if [[ $NEWREADINGORDER -eq 1 ]]
then
echo "Recalculating the Reading Order of the Regions..."
docker run $DOCKERGPUPARAMS -u $(id -u ${USER}):$(id -g ${USER}) --rm -it \
-v $IMAGES_PATH:$IMAGES_PATH/ \
-v $tmpdir:/tmp \
-v $(pwd):/workspace \
reorder-image \
bash -c "python /workspace/reorder/reorder.py $IMAGES_PATH/page --overwrite"
# Check if failed
status=$?
check_error_and_exit "Recalculating the Reading Order of the regions" $status
echo "Finished Recalculating the Reading Order of the regions."
fi
# cleanup results
rm -rf $tmpdir