-
Notifications
You must be signed in to change notification settings - Fork 257
/
Copy pathsuper_resolution_with_onnxruntime.html
915 lines (718 loc) ยท 69.2 KB
/
super_resolution_with_onnxruntime.html
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
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
<!DOCTYPE html>
<!--[if IE 8]><html class="no-js lt-ie9" lang="en" > <![endif]-->
<!--[if gt IE 8]><!--> <html class="no-js" lang="en" > <!--<![endif]-->
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>(์ ํ) PyTorch ๋ชจ๋ธ์ ONNX์ผ๋ก ๋ณํํ๊ณ ONNX ๋ฐํ์์์ ์คํํ๊ธฐ — PyTorch Tutorials 1.10.2+cu102 documentation</title>
<link rel="stylesheet" href="../_static/css/theme.css" type="text/css" />
<!-- <link rel="stylesheet" href="../_static/pygments.css" type="text/css" /> -->
<link rel="stylesheet" href="../_static/copybutton.css" type="text/css" />
<link rel="stylesheet" href="../_static/gallery.css" type="text/css" />
<link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/katex@0.10.0-beta/dist/katex.min.css" type="text/css" />
<link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/katex@0.13.11/dist/katex.min.css" type="text/css" />
<link rel="stylesheet" href="../_static/katex-math.css" type="text/css" />
<link rel="index" title="Index" href="../genindex.html" />
<link rel="search" title="Search" href="../search.html" />
<link rel="next" title="(๋ฒ ํ) FX์์ ํฉ์ฑ๊ณฑ/๋ฐฐ์น ์ ๊ทํ(Convolution/Batch Norm) ๊ฒฐํฉ๊ธฐ(Fuser) ๋ง๋ค๊ธฐ" href="../intermediate/fx_conv_bn_fuser.html" />
<link rel="prev" title="C++์์ TorchScript ๋ชจ๋ธ ๋ก๋ฉํ๊ธฐ" href="cpp_export.html" />
<script src="../_static/js/modernizr.min.js"></script>
<!-- Preload the theme fonts -->
<link rel="preload" href="../_static/fonts/FreightSans/freight-sans-book.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="../_static/fonts/FreightSans/freight-sans-medium.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="../_static/fonts/IBMPlexMono/IBMPlexMono-Medium.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="../_static/fonts/FreightSans/freight-sans-bold.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="../_static/fonts/FreightSans/freight-sans-medium-italic.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="../_static/fonts/IBMPlexMono/IBMPlexMono-SemiBold.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<!-- Preload the katex fonts -->
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Math-Italic.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Main-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Main-Bold.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Size1-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Size4-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Size2-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Size3-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="preload" href="https://cdn.jsdelivr.net/npm/katex@0.10.0/dist/fonts/KaTeX_Caligraphic-Regular.woff2" as="font" type="font/woff2" crossorigin="anonymous">
<link rel="stylesheet" href="https://use.fontawesome.com/releases/v5.15.2/css/all.css" integrity="sha384-vSIIfh2YWi9wW0r9iZe7RJPrKwp6bG+s9QZMoITbCckVJqGCCRhc+ccxNcdpHuYu" crossorigin="anonymous">
</head>
<div class="container-fluid header-holder tutorials-header" id="header-holder">
<div class="container">
<div class="header-container">
<a class="header-logo" href="https://pytorch.kr/" aria-label="PyTorch"></a>
<div class="main-menu">
<ul>
<li>
<a href="https://pytorch.kr/get-started">์์ํ๊ธฐ</a>
</li>
<li class="active">
<a href="https://tutorials.pytorch.kr">ํํ ๋ฆฌ์ผ</a>
</li>
<li>
<a href="https://pytorch.kr/hub">ํ๋ธ</a>
</li>
<li>
<a href="https://discuss.pytorch.kr">์ปค๋ฎค๋ํฐ</a>
</li>
</ul>
</div>
<a class="main-menu-open-button" href="#" data-behavior="open-mobile-menu"></a>
</div>
</div>
</div>
<body class="pytorch-body">
<div class="table-of-contents-link-wrapper">
<span>Table of Contents</span>
<a href="#" class="toggle-table-of-contents" data-behavior="toggle-table-of-contents"></a>
</div>
<nav data-toggle="wy-nav-shift" class="pytorch-left-menu" id="pytorch-left-menu">
<div class="pytorch-side-scroll">
<div class="pytorch-menu pytorch-menu-vertical" data-spy="affix" role="navigation" aria-label="main navigation">
<div class="pytorch-left-menu-search">
<div class="version">
1.10.2+cu102
</div>
<div role="search">
<form id="rtd-search-form" class="wy-form" action="../search.html" method="get">
<input type="text" name="q" placeholder="Search Tutorials" />
<input type="hidden" name="check_keywords" value="yes" />
<input type="hidden" name="area" value="default" />
</form>
</div>
</div>
<p class="caption"><span class="caption-text">ํ์ดํ ์น(PyTorch) ๋ ์ํผ</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../recipes/recipes_index.html">๋ชจ๋ ๋ ์ํผ ๋ณด๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../prototype/prototype_index.html">๋ชจ๋ ํ๋กํ ํ์
๋ ์ํผ ๋ณด๊ธฐ</a></li>
</ul>
<p class="caption"><span class="caption-text">ํ์ดํ ์น(PyTorch) ์์ํ๊ธฐ</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../beginner/basics/intro.html">ํ์ดํ ์น(PyTorch) ๊ธฐ๋ณธ ์ตํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/basics/quickstart_tutorial.html">๋น ๋ฅธ ์์(Quickstart)</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/basics/tensorqs_tutorial.html">ํ
์(Tensor)</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/basics/data_tutorial.html">Dataset๊ณผ DataLoader</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/basics/transforms_tutorial.html">๋ณํ(Transform)</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/basics/buildmodel_tutorial.html">์ ๊ฒฝ๋ง ๋ชจ๋ธ ๊ตฌ์ฑํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/basics/autogradqs_tutorial.html"><code class="docutils literal notranslate"><span class="pre">torch.autograd</span></code>๋ฅผ ์ฌ์ฉํ ์๋ ๋ฏธ๋ถ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/basics/optimization_tutorial.html">๋ชจ๋ธ ๋งค๊ฐ๋ณ์ ์ต์ ํํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/basics/saveloadrun_tutorial.html">๋ชจ๋ธ ์ ์ฅํ๊ณ ๋ถ๋ฌ์ค๊ธฐ</a></li>
</ul>
<p class="caption"><span class="caption-text">Introduction to PyTorch on YouTube</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../beginner/introyt.html">Introduction to PyTorch - YouTube Series</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/introyt/introyt1_tutorial.html">Introduction to PyTorch</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/introyt/tensors_deeper_tutorial.html">Introduction to PyTorch Tensors</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/introyt/autogradyt_tutorial.html">The Fundamentals of Autograd</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/introyt/modelsyt_tutorial.html">Building Models with PyTorch</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/introyt/tensorboardyt_tutorial.html">PyTorch TensorBoard Support</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/introyt/trainingyt.html">Training with PyTorch</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/introyt/captumyt.html">Model Understanding with Captum</a></li>
</ul>
<p class="caption"><span class="caption-text">ํ์ดํ ์น(PyTorch) ๋ฐฐ์ฐ๊ธฐ</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../beginner/deep_learning_60min_blitz.html">PyTorch๋ก ๋ฅ๋ฌ๋ํ๊ธฐ: 60๋ถ๋ง์ ๋์ฅ๋ด๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/pytorch_with_examples.html">์์ ๋ก ๋ฐฐ์ฐ๋ ํ์ดํ ์น(PyTorch)</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/nn_tutorial.html"><cite>torch.nn</cite> ์ด <em>์ค์ ๋ก</em> ๋ฌด์์ธ๊ฐ์?</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/tensorboard_tutorial.html">TensorBoard๋ก ๋ชจ๋ธ, ๋ฐ์ดํฐ, ํ์ต ์๊ฐํํ๊ธฐ</a></li>
</ul>
<p class="caption"><span class="caption-text">์ด๋ฏธ์ง/๋น๋์ค</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/torchvision_tutorial.html">TorchVision ๊ฐ์ฒด ๊ฒ์ถ ๋ฏธ์ธ์กฐ์ (Finetuning) ํํ ๋ฆฌ์ผ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/transfer_learning_tutorial.html">์ปดํจํฐ ๋น์ (Vision)์ ์ํ ์ ์ดํ์ต(Transfer Learning)</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/fgsm_tutorial.html">์ ๋์ ์์ ์์ฑ(Adversarial Example Generation)</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/dcgan_faces_tutorial.html">DCGAN ํํ ๋ฆฌ์ผ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/vt_tutorial.html">๋ฐฐํฌ๋ฅผ ์ํ ๋น์ ํธ๋์คํฌ๋จธ(Vision Transformer) ๋ชจ๋ธ ์ต์ ํํ๊ธฐ</a></li>
</ul>
<p class="caption"><span class="caption-text">์ค๋์ค</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../beginner/audio_io_tutorial.html">Audio I/O</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/audio_resampling_tutorial.html">Audio Resampling</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/audio_data_augmentation_tutorial.html">Audio Data Augmentation</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/audio_feature_extractions_tutorial.html">Audio Feature Extractions</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/audio_feature_augmentation_tutorial.html">Audio Feature Augmentation</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/audio_datasets_tutorial.html">Audio Datasets</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/speech_recognition_pipeline_tutorial.html">Speech Recognition with Wav2Vec2</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/speech_command_classification_with_torchaudio_tutorial.html">Speech Command Classification with torchaudio</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/text_to_speech_with_torchaudio.html">Text-to-speech with torchaudio</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/forced_alignment_with_torchaudio_tutorial.html">Forced Alignment with Wav2Vec2</a></li>
</ul>
<p class="caption"><span class="caption-text">ํ
์คํธ</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../beginner/transformer_tutorial.html">nn.Transformer ์ TorchText ๋ก ์ํ์ค-ํฌ-์ํ์ค(Sequence-to-Sequence) ๋ชจ๋ธ๋งํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/char_rnn_classification_tutorial.html">๊ธฐ์ด๋ถํฐ ์์ํ๋ NLP: ๋ฌธ์-๋จ์ RNN์ผ๋ก ์ด๋ฆ ๋ถ๋ฅํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/char_rnn_generation_tutorial.html">๊ธฐ์ด๋ถํฐ ์์ํ๋ NLP: ๋ฌธ์-๋จ์ RNN์ผ๋ก ์ด๋ฆ ์์ฑํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/seq2seq_translation_tutorial.html">๊ธฐ์ด๋ถํฐ ์์ํ๋ NLP: Sequence to Sequence ๋คํธ์ํฌ์ Attention์ ์ด์ฉํ ๋ฒ์ญ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/text_sentiment_ngrams_tutorial.html">torchtext ๋ผ์ด๋ธ๋ฌ๋ฆฌ๋ก ํ
์คํธ ๋ถ๋ฅํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/translation_transformer.html">nn.Transformer์ torchtext๋ก ์ธ์ด ๋ฒ์ญํ๊ธฐ</a></li>
</ul>
<p class="caption"><span class="caption-text">๊ฐํํ์ต</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/reinforcement_q_learning.html">๊ฐํ ํ์ต (DQN) ํํ ๋ฆฌ์ผ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/mario_rl_tutorial.html">Train a Mario-playing RL Agent</a></li>
</ul>
<p class="caption"><span class="caption-text">PyTorch ๋ชจ๋ธ์ ํ๋ก๋์
ํ๊ฒฝ์ ๋ฐฐํฌํ๊ธฐ</span></p>
<ul class="current">
<li class="toctree-l1"><a class="reference internal" href="../intermediate/flask_rest_api_tutorial.html">Flask๋ฅผ ์ฌ์ฉํ์ฌ Python์์ PyTorch๋ฅผ REST API๋ก ๋ฐฐํฌํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/Intro_to_TorchScript_tutorial.html">TorchScript ์๊ฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="cpp_export.html">C++์์ TorchScript ๋ชจ๋ธ ๋ก๋ฉํ๊ธฐ</a></li>
<li class="toctree-l1 current"><a class="current reference internal" href="#">(์ ํ) PyTorch ๋ชจ๋ธ์ ONNX์ผ๋ก ๋ณํํ๊ณ ONNX ๋ฐํ์์์ ์คํํ๊ธฐ</a></li>
</ul>
<p class="caption"><span class="caption-text">Code Transforms with FX</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/fx_conv_bn_fuser.html">(๋ฒ ํ) FX์์ ํฉ์ฑ๊ณฑ/๋ฐฐ์น ์ ๊ทํ(Convolution/Batch Norm) ๊ฒฐํฉ๊ธฐ(Fuser) ๋ง๋ค๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/fx_profiling_tutorial.html">(beta) Building a Simple CPU Performance Profiler with FX</a></li>
</ul>
<p class="caption"><span class="caption-text">ํ๋ก ํธ์๋ API</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/memory_format_tutorial.html">(๋ฒ ํ) PyTorch๋ฅผ ์ฌ์ฉํ Channels Last ๋ฉ๋ชจ๋ฆฌ ํ์</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/forward_ad_usage.html">Forward-mode Automatic Differentiation (Beta)</a></li>
<li class="toctree-l1"><a class="reference internal" href="cpp_frontend.html">PyTorch C++ ํ๋ก ํธ์๋ ์ฌ์ฉํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="torch-script-parallelism.html">TorchScript์ ๋์ ๋ณ๋ ฌ ์ฒ๋ฆฌ(Dynamic Parallelism)</a></li>
<li class="toctree-l1"><a class="reference internal" href="cpp_autograd.html">C++ ํ๋ก ํธ์๋์ ์๋ ๋ฏธ๋ถ (autograd)</a></li>
</ul>
<p class="caption"><span class="caption-text">PyTorch ํ์ฅํ๊ธฐ</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/custom_function_double_backward_tutorial.html">Double Backward with Custom Functions</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/custom_function_conv_bn_tutorial.html">Fusing Convolution and Batch Norm using Custom Function</a></li>
<li class="toctree-l1"><a class="reference internal" href="cpp_extension.html">Custom C++ and CUDA Extensions</a></li>
<li class="toctree-l1"><a class="reference internal" href="torch_script_custom_ops.html">Extending TorchScript with Custom C++ Operators</a></li>
<li class="toctree-l1"><a class="reference internal" href="torch_script_custom_classes.html">์ปค์คํ
C++ ํด๋์ค๋ก TorchScript ํ์ฅํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="dispatcher.html">Registering a Dispatched Operator in C++</a></li>
<li class="toctree-l1"><a class="reference internal" href="extend_dispatcher.html">Extending dispatcher for a new backend in C++</a></li>
</ul>
<p class="caption"><span class="caption-text">๋ชจ๋ธ ์ต์ ํ</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../beginner/profiler.html">PyTorch ๋ชจ๋ ํ๋กํ์ผ๋ง ํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/tensorboard_profiler_tutorial.html">PyTorch Profiler With TensorBoard</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/hyperparameter_tuning_tutorial.html">Hyperparameter tuning with Ray Tune</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/vt_tutorial.html">๋ฐฐํฌ๋ฅผ ์ํ ๋น์ ํธ๋์คํฌ๋จธ(Vision Transformer) ๋ชจ๋ธ ์ต์ ํํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/parametrizations.html">Parametrizations Tutorial</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/pruning_tutorial.html">๊ฐ์ง์น๊ธฐ ๊ธฐ๋ฒ(Pruning) ํํ ๋ฆฌ์ผ</a></li>
<li class="toctree-l1"><a class="reference internal" href="dynamic_quantization_tutorial.html">(๋ฒ ํ) LSTM ๊ธฐ๋ฐ ๋จ์ด ๋จ์ ์ธ์ด ๋ชจ๋ธ์ ๋์ ์์ํ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/dynamic_quantization_bert_tutorial.html">(๋ฒ ํ) BERT ๋ชจ๋ธ ๋์ ์์ํํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/quantized_transfer_learning_tutorial.html">(๋ฒ ํ) ์ปดํจํฐ ๋น์ ํํ ๋ฆฌ์ผ์ ์ํ ์์ํ๋ ์ ์ดํ์ต(Quantized Transfer Learning)</a></li>
<li class="toctree-l1"><a class="reference internal" href="static_quantization_tutorial.html">(beta) Static Quantization with Eager Mode in PyTorch</a></li>
</ul>
<p class="caption"><span class="caption-text">๋ณ๋ ฌ ๋ฐ ๋ถ์ฐ ํ์ต</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../beginner/dist_overview.html">PyTorch Distributed Overview</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/model_parallel_tutorial.html">๋จ์ผ ๋จธ์ ์ ์ฌ์ฉํ ๋ชจ๋ธ ๋ณ๋ ฌํ ๋ชจ๋ฒ ์ฌ๋ก</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/ddp_tutorial.html">๋ถ์ฐ ๋ฐ์ดํฐ ๋ณ๋ ฌ ์ฒ๋ฆฌ ์์ํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/dist_tuto.html">PyTorch๋ก ๋ถ์ฐ ์ดํ๋ฆฌ์ผ์ด์
๊ฐ๋ฐํ๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/rpc_tutorial.html">Getting Started with Distributed RPC Framework</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/rpc_param_server_tutorial.html">Implementing a Parameter Server Using Distributed RPC Framework</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/dist_pipeline_parallel_tutorial.html">Distributed Pipeline Parallelism Using RPC</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/rpc_async_execution.html">Implementing Batch RPC Processing Using Asynchronous Executions</a></li>
<li class="toctree-l1"><a class="reference internal" href="rpc_ddp_tutorial.html">๋ถ์ฐ ๋ฐ์ดํฐ ๋ณ๋ ฌ(DDP)๊ณผ ๋ถ์ฐ RPC ํ๋ ์์ํฌ ๊ฒฐํฉ</a></li>
<li class="toctree-l1"><a class="reference internal" href="../intermediate/pipeline_tutorial.html">ํ์ดํ๋ผ์ธ ๋ณ๋ ฌํ๋ก ํธ๋์คํฌ๋จธ ๋ชจ๋ธ ํ์ต์ํค๊ธฐ</a></li>
<li class="toctree-l1"><a class="reference internal" href="ddp_pipeline.html">๋ถ์ฐ ๋ฐ์ดํฐ ๋ณ๋ ฌ ์ฒ๋ฆฌ์ ๋ณ๋ ฌ ์ฒ๋ฆฌ ํ์ดํ๋ผ์ธ์ ์ฌ์ฉํ ํธ๋์คํฌ๋จธ ๋ชจ๋ธ ํ์ต</a></li>
<li class="toctree-l1"><a class="reference internal" href="generic_join.html">Distributed Training with Uneven Inputs Using the Join Context Manager</a></li>
</ul>
<p class="caption"><span class="caption-text">Mobile</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../beginner/deeplabv3_on_ios.html">iOS์์์ ์ด๋ฏธ์ง ๋ถํ DeepLapV3</a></li>
<li class="toctree-l1"><a class="reference internal" href="../beginner/deeplabv3_on_android.html">์๋๋ก์ด๋์์์ ์ด๋ฏธ์ง ๋ถํ DeepLapV3</a></li>
</ul>
</div>
</div>
</nav>
<div class="pytorch-container">
<div class="pytorch-page-level-bar" id="pytorch-page-level-bar">
<div class="pytorch-breadcrumbs-wrapper">
<div role="navigation" aria-label="breadcrumbs navigation">
<ul class="pytorch-breadcrumbs">
<li>
<a href="../index.html">
Tutorials
</a> >
</li>
<li>(์ ํ) PyTorch ๋ชจ๋ธ์ ONNX์ผ๋ก ๋ณํํ๊ณ ONNX ๋ฐํ์์์ ์คํํ๊ธฐ</li>
<li class="pytorch-breadcrumbs-aside">
<a href="../_sources/advanced/super_resolution_with_onnxruntime.rst.txt" rel="nofollow"><img src="../_static/images/view-page-source-icon.svg"></a>
</li>
</ul>
</div>
</div>
<div class="pytorch-shortcuts-wrapper" id="pytorch-shortcuts-wrapper">
Shortcuts
</div>
</div>
<section data-toggle="wy-nav-shift" id="pytorch-content-wrap" class="pytorch-content-wrap">
<div class="pytorch-content-left">
<div class="pytorch-call-to-action-links">
<div id="tutorial-type">advanced/super_resolution_with_onnxruntime</div>
<div id="google-colab-link">
<img class="call-to-action-img" src="../_static/images/pytorch-colab.svg"/>
<div class="call-to-action-desktop-view">Run in Google Colab</div>
<div class="call-to-action-mobile-view">Colab</div>
</div>
<div id="download-notebook-link">
<img class="call-to-action-notebook-img" src="../_static/images/pytorch-download.svg"/>
<div class="call-to-action-desktop-view">Download Notebook</div>
<div class="call-to-action-mobile-view">Notebook</div>
</div>
<div id="github-view-link">
<img class="call-to-action-img" src="../_static/images/pytorch-github.svg"/>
<div class="call-to-action-desktop-view">View on GitHub</div>
<div class="call-to-action-mobile-view">GitHub</div>
</div>
</div>
<div class="rst-content">
<div role="main" class="main-content" itemscope="itemscope" itemtype="http://schema.org/Article">
<article itemprop="articleBody" id="pytorch-article" class="pytorch-article">
<div class="sphx-glr-download-link-note admonition note">
<p class="first admonition-title">Note</p>
<p class="last">Click <a class="reference internal" href="#sphx-glr-download-advanced-super-resolution-with-onnxruntime-py"><span class="std std-ref">here</span></a> to download the full example code</p>
</div>
<div class="sphx-glr-example-title section" id="pytorch-onnx-onnx">
<span id="sphx-glr-advanced-super-resolution-with-onnxruntime-py"></span><h1>(์ ํ) PyTorch ๋ชจ๋ธ์ ONNX์ผ๋ก ๋ณํํ๊ณ ONNX ๋ฐํ์์์ ์คํํ๊ธฐ<a class="headerlink" href="#pytorch-onnx-onnx" title="Permalink to this headline">ยถ</a></h1>
<p>์ด ํํ ๋ฆฌ์ผ์์๋ ์ด๋ป๊ฒ PyTorch์์ ์ ์๋ ๋ชจ๋ธ์ ONNX ํ์์ผ๋ก ๋ณํํ๊ณ ๋ ์ด๋ป๊ฒ ๊ทธ ๋ณํ๋ ๋ชจ๋ธ์
ONNX ๋ฐํ์์์ ์คํํ ์ ์๋์ง์ ๋ํด ์์๋ณด๋๋ก ํ๊ฒ ์ต๋๋ค.
ONNX ๋ฐํ์์ ONNX ๋ชจ๋ธ์ ์ํ ์์ง์ผ๋ก์ ์ฑ๋ฅ์ ์ด์ ์ ๋ง์ถ๊ณ ์๊ณ ์ฌ๋ฌ ๋ค์ํ ํ๋ซํผ๊ณผ ํ๋์จ์ด(์๋์ฐ,
๋ฆฌ๋
์ค, ๋งฅ์ ๋น๋กฏํ ํ๋ซํผ ๋ฟ๋ง ์๋๋ผ CPU, GPU ๋ฑ์ ํ๋์จ์ด)์์ ํจ์จ์ ์ธ ์ถ๋ก ์ ๊ฐ๋ฅํ๊ฒ ํฉ๋๋ค.
ONNX ๋ฐํ์์ <a class="reference external" href="https://cloudblogs.microsoft.com/opensource/2019/05/22/onnx-runtime-machine-learning-inferencing-0-4-release">์ฌ๊ธฐ</a> ์์
์ค๋ช
๋ ๊ฒ๊ณผ ๊ฐ์ด ์ฌ๋ฌ ๋ชจ๋ธ๋ค์ ์ฑ๋ฅ์ ์๋นํ ๋์ผ ์ ์๋ค๋ ์ ์ด ์ฆ๋ช
๋์์ต๋๋ค.
์ด ํํ ๋ฆฌ์ผ์ ์งํํ๊ธฐ ์ํด์๋ <a class="reference external" href="https://github.com/onnx/onnx">ONNX</a>
์ <a class="reference external" href="https://github.com/microsoft/onnxruntime">ONNX Runtime</a> ์ค์น๊ฐ ํ์ํฉ๋๋ค.
ONNX์ ONNX ๋ฐํ์์ ๋ฐ์ด๋๋ฆฌ ๋น๋๋ฅผ <code class="docutils literal notranslate"><span class="pre">pip</span> <span class="pre">install</span> <span class="pre">onnx</span> <span class="pre">onnxruntime</span></code> ๋ฅผ ํตํด ๋ฐ์ ์ ์์ต๋๋ค.
ONNX ๋ฐํ์์ ๋ฒ์ 3.5์์ 3.7๊น์ง์ Python๊ณผ ํธํ๋ฉ๋๋ค.
<code class="docutils literal notranslate"><span class="pre">์ฐธ๊ณ </span></code>: ๋ณธ ํํ ๋ฆฌ์ผ์ PyTorch์ master ๋ธ๋์น๋ฅผ ํ์๋กํ๋ฉฐ <a class="reference external" href="https://github.com/pytorch/pytorch#from-source">๋งํฌ</a> ์์
์ค์นํ ์ ์์ต๋๋ค.</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="c1"># ํ์ํ import๋ฌธ</span>
<span class="kn">import</span> <span class="nn">io</span>
<span class="kn">import</span> <span class="nn">numpy</span> <span class="k">as</span> <span class="nn">np</span>
<span class="kn">from</span> <span class="nn">torch</span> <span class="kn">import</span> <span class="n">nn</span>
<span class="kn">import</span> <span class="nn">torch.utils.model_zoo</span> <span class="k">as</span> <span class="nn">model_zoo</span>
<span class="kn">import</span> <span class="nn">torch.onnx</span>
</pre></div>
</div>
<p>์ดํด์ํ(super-resolution)๋ ์ด๋ฏธ์ง๋ ๋น๋์ค์ ํด์๋๋ฅผ ๋์ด๊ธฐ ์ํ ๋ฐฉ๋ฒ์ผ๋ก ์ด๋ฏธ์ง ํ๋ก์ธ์ฑ์ด๋ ๋น๋์ค ํธ์ง์ ๋๋ฆฌ
์ฌ์ฉ๋๊ณ ์๋ ๋ฐฉ๋ฒ์
๋๋ค. ์ด ํํ ๋ฆฌ์ผ์์๋ ํฌ๊ธฐ๊ฐ ์์ ์ดํด์ํ ๋ชจ๋ธ์ ์ฌ์ฉํ๋๋ก ํ๊ฒ ์ต๋๋ค.</p>
<p>๋จผ์ , ์ดํด์ํ ๋ชจ๋ธ์ PyTorch์์ ๊ตฌํํ๊ฒ ์ต๋๋ค.
์ด ๋ชจ๋ธ์ <a class="reference external" href="https://arxiv.org/abs/1609.05158">โReal-Time Single Image and Video Super-Resolution Using an Efficient
Sub-Pixel Convolutional Neural Networkโ - Shi et al</a> ์์ ์๊ฐ๋
ํจ์จ์ ์ธ ์๋ธํฝ์
ํฉ์ฑ๊ณฑ ๊ณ์ธต์ ์ฌ์ฉํ์ฌ ์ด๋ฏธ์ง์ ํด์๋๋ฅผ ์
์ค์ผ์ผ ์ธ์๋งํผ ๋๋ฆฝ๋๋ค.
๋ชจ๋ธ์ ์ด๋ฏธ์ง์ YCbCr ์ฑ๋ถ ์ค Y ์ฑ๋ถ์ ์
๋ ฅ๊ฐ์ผ๋ก ๋ฐ๊ณ ์
์ค์ผ์ผ๋ ์ดํด์๋์ Y ์ฑ๋ ๊ฐ์ ๋ฆฌํดํฉ๋๋ค.</p>
<p>์๋๋ PyTorch ์์ ์
<a class="reference external" href="https://github.com/pytorch/examples/blob/master/super_resolution/model.py">๋ชจ๋ธ</a>
์ ๊ทธ๋๋ก ๊ฐ์ ธ์จ ๊ฒ์
๋๋ค:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="c1"># PyTorch์์ ๊ตฌํ๋ ์ดํด์๋ ๋ชจ๋ธ</span>
<span class="kn">import</span> <span class="nn">torch.nn</span> <span class="k">as</span> <span class="nn">nn</span>
<span class="kn">import</span> <span class="nn">torch.nn.init</span> <span class="k">as</span> <span class="nn">init</span>
<span class="k">class</span> <span class="nc">SuperResolutionNet</span><span class="p">(</span><span class="n">nn</span><span class="o">.</span><span class="n">Module</span><span class="p">):</span>
<span class="k">def</span> <span class="fm">__init__</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">upscale_factor</span><span class="p">,</span> <span class="n">inplace</span><span class="o">=</span><span class="kc">False</span><span class="p">):</span>
<span class="nb">super</span><span class="p">(</span><span class="n">SuperResolutionNet</span><span class="p">,</span> <span class="bp">self</span><span class="p">)</span><span class="o">.</span><span class="fm">__init__</span><span class="p">()</span>
<span class="bp">self</span><span class="o">.</span><span class="n">relu</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">ReLU</span><span class="p">(</span><span class="n">inplace</span><span class="o">=</span><span class="n">inplace</span><span class="p">)</span>
<span class="bp">self</span><span class="o">.</span><span class="n">conv1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Conv2d</span><span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="mi">64</span><span class="p">,</span> <span class="p">(</span><span class="mi">5</span><span class="p">,</span> <span class="mi">5</span><span class="p">),</span> <span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="mi">1</span><span class="p">),</span> <span class="p">(</span><span class="mi">2</span><span class="p">,</span> <span class="mi">2</span><span class="p">))</span>
<span class="bp">self</span><span class="o">.</span><span class="n">conv2</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Conv2d</span><span class="p">(</span><span class="mi">64</span><span class="p">,</span> <span class="mi">64</span><span class="p">,</span> <span class="p">(</span><span class="mi">3</span><span class="p">,</span> <span class="mi">3</span><span class="p">),</span> <span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="mi">1</span><span class="p">),</span> <span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="mi">1</span><span class="p">))</span>
<span class="bp">self</span><span class="o">.</span><span class="n">conv3</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Conv2d</span><span class="p">(</span><span class="mi">64</span><span class="p">,</span> <span class="mi">32</span><span class="p">,</span> <span class="p">(</span><span class="mi">3</span><span class="p">,</span> <span class="mi">3</span><span class="p">),</span> <span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="mi">1</span><span class="p">),</span> <span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="mi">1</span><span class="p">))</span>
<span class="bp">self</span><span class="o">.</span><span class="n">conv4</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Conv2d</span><span class="p">(</span><span class="mi">32</span><span class="p">,</span> <span class="n">upscale_factor</span> <span class="o">**</span> <span class="mi">2</span><span class="p">,</span> <span class="p">(</span><span class="mi">3</span><span class="p">,</span> <span class="mi">3</span><span class="p">),</span> <span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="mi">1</span><span class="p">),</span> <span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="mi">1</span><span class="p">))</span>
<span class="bp">self</span><span class="o">.</span><span class="n">pixel_shuffle</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">PixelShuffle</span><span class="p">(</span><span class="n">upscale_factor</span><span class="p">)</span>
<span class="bp">self</span><span class="o">.</span><span class="n">_initialize_weights</span><span class="p">()</span>
<span class="k">def</span> <span class="nf">forward</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">x</span><span class="p">):</span>
<span class="n">x</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">relu</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">conv1</span><span class="p">(</span><span class="n">x</span><span class="p">))</span>
<span class="n">x</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">relu</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">conv2</span><span class="p">(</span><span class="n">x</span><span class="p">))</span>
<span class="n">x</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">relu</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">conv3</span><span class="p">(</span><span class="n">x</span><span class="p">))</span>
<span class="n">x</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">pixel_shuffle</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">conv4</span><span class="p">(</span><span class="n">x</span><span class="p">))</span>
<span class="k">return</span> <span class="n">x</span>
<span class="k">def</span> <span class="nf">_initialize_weights</span><span class="p">(</span><span class="bp">self</span><span class="p">):</span>
<span class="n">init</span><span class="o">.</span><span class="n">orthogonal_</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">conv1</span><span class="o">.</span><span class="n">weight</span><span class="p">,</span> <span class="n">init</span><span class="o">.</span><span class="n">calculate_gain</span><span class="p">(</span><span class="s1">'relu'</span><span class="p">))</span>
<span class="n">init</span><span class="o">.</span><span class="n">orthogonal_</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">conv2</span><span class="o">.</span><span class="n">weight</span><span class="p">,</span> <span class="n">init</span><span class="o">.</span><span class="n">calculate_gain</span><span class="p">(</span><span class="s1">'relu'</span><span class="p">))</span>
<span class="n">init</span><span class="o">.</span><span class="n">orthogonal_</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">conv3</span><span class="o">.</span><span class="n">weight</span><span class="p">,</span> <span class="n">init</span><span class="o">.</span><span class="n">calculate_gain</span><span class="p">(</span><span class="s1">'relu'</span><span class="p">))</span>
<span class="n">init</span><span class="o">.</span><span class="n">orthogonal_</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">conv4</span><span class="o">.</span><span class="n">weight</span><span class="p">)</span>
<span class="c1"># ์์์ ์ ์๋ ๋ชจ๋ธ์ ์ฌ์ฉํ์ฌ ์ดํด์๋ ๋ชจ๋ธ ์์ฑ</span>
<span class="n">torch_model</span> <span class="o">=</span> <span class="n">SuperResolutionNet</span><span class="p">(</span><span class="n">upscale_factor</span><span class="o">=</span><span class="mi">3</span><span class="p">)</span>
</pre></div>
</div>
<p>์ ๊ณผ์ ์ด ๋๋๋ฉด ์ผ๋ฐ์ ์ธ ๊ฒฝ์ฐ์ ๋ชจ๋ธ์ ํ์ต์ํค๊ธฐ ์์ํ ๊ฒ์
๋๋ค. ํ์ง๋ง ๋ณธ ํํ ๋ฆฌ์ผ์์๋
๋ฏธ๋ฆฌ ํ์ต๋ ๊ฐ์ค์น๋ค์ ์ฌ์ฉํ๋๋ก ํ๊ฒ ์ต๋๋ค. ์ฐธ๊ณ ๋ก ์ด ๋ชจ๋ธ์ ๋์ ์ ํ๋์ ์ด๋ฅผ ๋๊น์ง ํ์ต๋์ง ์์๊ณ
๋ณธ ํํ ๋ฆฌ์ผ์ ์ํํ ์งํํ๊ธฐ ์ํ ๋ชฉ์ ์ผ๋ก ์ฌ์ฉํ๋ ๊ฒ์
๋๋ค.</p>
<p>๋ชจ๋ธ์ ๋ณํํ๊ธฐ ์ ์ ๋ชจ๋ธ์ ์ถ๋ก ๋ชจ๋๋ก ๋ฐ๊พธ๊ธฐ ์ํด์ <code class="docutils literal notranslate"><span class="pre">torch_model.eval()</span></code> ๋๋ <code class="docutils literal notranslate"><span class="pre">torch_model.train(False)</span></code> ๋ฅผ
ํธ์ถํ๋ ๊ฒ์ด ์ค์ํฉ๋๋ค.
์ด๋ dropout์ด๋ batchnorm๊ณผ ๊ฐ์ ์ฐ์ฐ๋ค์ด ์ถ๋ก ๊ณผ ํ์ต ๋ชจ๋์์ ๋ค๋ฅด๊ฒ ์๋ํ๊ธฐ ๋๋ฌธ์ ํ์ํฉ๋๋ค.</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="c1"># ๋ฏธ๋ฆฌ ํ์ต๋ ๊ฐ์ค์น๋ฅผ ์ฝ์ด์ต๋๋ค</span>
<span class="n">model_url</span> <span class="o">=</span> <span class="s1">'https://s3.amazonaws.com/pytorch/test_data/export/superres_epoch100-44c6958e.pth'</span>
<span class="n">batch_size</span> <span class="o">=</span> <span class="mi">1</span> <span class="c1"># ์์์ ์</span>
<span class="c1"># ๋ชจ๋ธ์ ๋ฏธ๋ฆฌ ํ์ต๋ ๊ฐ์ค์น๋ก ์ด๊ธฐํํฉ๋๋ค</span>
<span class="n">map_location</span> <span class="o">=</span> <span class="k">lambda</span> <span class="n">storage</span><span class="p">,</span> <span class="n">loc</span><span class="p">:</span> <span class="n">storage</span>
<span class="k">if</span> <span class="n">torch</span><span class="o">.</span><span class="n">cuda</span><span class="o">.</span><span class="n">is_available</span><span class="p">():</span>
<span class="n">map_location</span> <span class="o">=</span> <span class="kc">None</span>
<span class="n">torch_model</span><span class="o">.</span><span class="n">load_state_dict</span><span class="p">(</span><span class="n">model_zoo</span><span class="o">.</span><span class="n">load_url</span><span class="p">(</span><span class="n">model_url</span><span class="p">,</span> <span class="n">map_location</span><span class="o">=</span><span class="n">map_location</span><span class="p">))</span>
<span class="c1"># ๋ชจ๋ธ์ ์ถ๋ก ๋ชจ๋๋ก ์ ํํฉ๋๋ค</span>
<span class="n">torch_model</span><span class="o">.</span><span class="n">eval</span><span class="p">()</span>
</pre></div>
</div>
<p>์ด์ Tracing์ด๋ ์คํฌ๋ฆฝํ
์ ํตํด์ PyTorch ๋ชจ๋ธ์ ๋ณํํ ์ ์์ต๋๋ค.
์ด ํํ ๋ฆฌ์ผ์์๋ tracing์ ํตํด ๋ณํ๋ ๋ชจ๋ธ์ ์ฌ์ฉํ๋๋ก ํ๊ฒ ์ต๋๋ค.
๋ชจ๋ธ์ ๋ณํํ๊ธฐ ์ํด์๋ <code class="docutils literal notranslate"><span class="pre">torch.onnx.export()</span></code> ํจ์๋ฅผ ํธ์ถํฉ๋๋ค.
์ด ํจ์๋ ๋ชจ๋ธ์ ์คํํ์ฌ ์ด๋ค ์ฐ์ฐ์๋ค์ด ์ถ๋ ฅ๊ฐ์ ๊ณ์ฐํ๋๋ฐ ์ฌ์ฉ๋์๋์ง๋ฅผ ๊ธฐ๋กํฉ๋๋ค.
<code class="docutils literal notranslate"><span class="pre">export</span></code> ํจ์๊ฐ ๋ชจ๋ธ์ ์คํํ๊ธฐ ๋๋ฌธ์, ์ฐ๋ฆฌ๊ฐ ์ง์ ํ
์๋ฅผ ์
๋ ฅ๊ฐ์ผ๋ก ๋๊ฒจ์ฃผ์ด์ผ ํฉ๋๋ค.
์ด ํ
์์ ๊ฐ์ ์๋ง์ ์๋ฃํ๊ณผ ๋ชจ์์ด๋ผ๋ฉด ๋๋คํ๊ฒ ๊ฒฐ์ ๋์ด๋ ๋ฌด๋ฐฉํฉ๋๋ค.
ํน์ ์ฐจ์์ ๋์ ์ฐจ์์ผ๋ก ์ง์ ํ์ง ์๋ ์ด์, ONNX๋ก ๋ณํ๋ ๊ทธ๋ํ์ ๊ฒฝ์ฐ ์
๋ ฅ๊ฐ์ ํฌ๊ธฐ๋ ๋ชจ๋ ์ฐจ์์ ๋ํด ๊ณ ์ ๋ฉ๋๋ค.
์์์์๋ ๋ชจ๋ธ์ด ํญ์ ๋ฐฐ์น ์ฌ์ด์ฆ 1์ ์ฌ์ฉํ๋๋ก ๋ณํํ์์ง๋ง, <code class="docutils literal notranslate"><span class="pre">torch.onnx.export()</span></code> ์ <code class="docutils literal notranslate"><span class="pre">dynamic_axes</span></code> ์ธ์์
์ฒซ๋ฒ์งธ ์ฐจ์์ ๋์ ์ฐจ์์ผ๋ก ์ง์ ํฉ๋๋ค. ๋ฐ๋ผ์ ๋ณํ๋ ๋ชจ๋ธ์ ์์์ batch_size์ ๋ํด [batch_size, 1, 224, 224] ์ฌ์ด์ฆ
์
๋ ฅ๊ฐ์ ๋ฐ์ ์ ์์ต๋๋ค.</p>
<p>PyTorch์ ๋ณํ ์ธํฐํ์ด์ค์ ๋ํด ๋ ์์ธํ ์๊ณ ์ถ๋ค๋ฉด
<a class="reference external" href="https://pytorch.org/docs/master/onnx.html">torch.onnx ๋ฌธ์</a> ๋ฅผ ์ฐธ๊ณ ํด์ฃผ์ธ์.</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="c1"># ๋ชจ๋ธ์ ๋ํ ์
๋ ฅ๊ฐ</span>
<span class="n">x</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">randn</span><span class="p">(</span><span class="n">batch_size</span><span class="p">,</span> <span class="mi">1</span><span class="p">,</span> <span class="mi">224</span><span class="p">,</span> <span class="mi">224</span><span class="p">,</span> <span class="n">requires_grad</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span>
<span class="n">torch_out</span> <span class="o">=</span> <span class="n">torch_model</span><span class="p">(</span><span class="n">x</span><span class="p">)</span>
<span class="c1"># ๋ชจ๋ธ ๋ณํ</span>
<span class="n">torch</span><span class="o">.</span><span class="n">onnx</span><span class="o">.</span><span class="n">export</span><span class="p">(</span><span class="n">torch_model</span><span class="p">,</span> <span class="c1"># ์คํ๋ ๋ชจ๋ธ</span>
<span class="n">x</span><span class="p">,</span> <span class="c1"># ๋ชจ๋ธ ์
๋ ฅ๊ฐ (ํํ ๋๋ ์ฌ๋ฌ ์
๋ ฅ๊ฐ๋ค๋ ๊ฐ๋ฅ)</span>
<span class="s2">"super_resolution.onnx"</span><span class="p">,</span> <span class="c1"># ๋ชจ๋ธ ์ ์ฅ ๊ฒฝ๋ก (ํ์ผ ๋๋ ํ์ผ๊ณผ ์ ์ฌํ ๊ฐ์ฒด ๋ชจ๋ ๊ฐ๋ฅ)</span>
<span class="n">export_params</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span> <span class="c1"># ๋ชจ๋ธ ํ์ผ ์์ ํ์ต๋ ๋ชจ๋ธ ๊ฐ์ค์น๋ฅผ ์ ์ฅํ ์ง์ ์ฌ๋ถ</span>
<span class="n">opset_version</span><span class="o">=</span><span class="mi">10</span><span class="p">,</span> <span class="c1"># ๋ชจ๋ธ์ ๋ณํํ ๋ ์ฌ์ฉํ ONNX ๋ฒ์ </span>
<span class="n">do_constant_folding</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span> <span class="c1"># ์ต์ ํ์ ์์ํด๋ฉ์ ์ฌ์ฉํ ์ง์ ์ฌ๋ถ</span>
<span class="n">input_names</span> <span class="o">=</span> <span class="p">[</span><span class="s1">'input'</span><span class="p">],</span> <span class="c1"># ๋ชจ๋ธ์ ์
๋ ฅ๊ฐ์ ๊ฐ๋ฆฌํค๋ ์ด๋ฆ</span>
<span class="n">output_names</span> <span class="o">=</span> <span class="p">[</span><span class="s1">'output'</span><span class="p">],</span> <span class="c1"># ๋ชจ๋ธ์ ์ถ๋ ฅ๊ฐ์ ๊ฐ๋ฆฌํค๋ ์ด๋ฆ</span>
<span class="n">dynamic_axes</span><span class="o">=</span><span class="p">{</span><span class="s1">'input'</span> <span class="p">:</span> <span class="p">{</span><span class="mi">0</span> <span class="p">:</span> <span class="s1">'batch_size'</span><span class="p">},</span> <span class="c1"># ๊ฐ๋ณ์ ์ธ ๊ธธ์ด๋ฅผ ๊ฐ์ง ์ฐจ์</span>
<span class="s1">'output'</span> <span class="p">:</span> <span class="p">{</span><span class="mi">0</span> <span class="p">:</span> <span class="s1">'batch_size'</span><span class="p">}})</span>
</pre></div>
</div>
<p>ONNX ๋ฐํ์์์ ๋ณํ๋ ๋ชจ๋ธ์ ์ฌ์ฉํ์ ๋ ๊ฐ์ ๊ฒฐ๊ณผ๋ฅผ ์ป๋์ง ํ์ธํ๊ธฐ ์ํด์ <code class="docutils literal notranslate"><span class="pre">torch_out</span></code> ๋ฅผ ๊ณ์ฐํฉ๋๋ค.</p>
<p>ONNX ๋ฐํ์์์์ ๋ชจ๋ธ ๊ฒฐ๊ณผ๊ฐ์ ํ์ธํ๊ธฐ ์ ์ ๋จผ์ ONNX API๋ฅผ ์ฌ์ฉํด ONNX ๋ชจ๋ธ์ ํ์ธํด๋ณด๋๋ก ํ๊ฒ ์ต๋๋ค.
๋จผ์ , <code class="docutils literal notranslate"><span class="pre">onnx.load("super_resolution.onnx")</span></code> ๋ ์ ์ฅ๋ ๋ชจ๋ธ์ ์ฝ์ด์จ ํ
๋จธ์ ๋ฌ๋ ๋ชจ๋ธ์ ์ทจํฉํ์ฌ ์ ์ฅํ๊ณ ์๋ ์์ ํ์ผ ์ปจํ
์ด๋์ธ onnx.ModelProto๋ฅผ ๋ฆฌํดํฉ๋๋ค.
onnx.ModelProto์ ๋ํด ๋ ์์ธํ ๊ฒ์ <a class="reference external" href="https://github.com/onnx/onnx/blob/master/onnx/onnx.proto">onnx.proto ๊ธฐ์ ๋ฌธ์</a> ์์
ํ์ธํ์ค ์ ์์ต๋๋ค.
<code class="docutils literal notranslate"><span class="pre">onnx.checker.check_model(onnx_model)</span></code> ๋ ๋ชจ๋ธ์ ๊ตฌ์กฐ๋ฅผ ํ์ธํ๊ณ
๋ชจ๋ธ์ด ์ ํจํ ์คํค๋ง(valid schema)๋ฅผ ๊ฐ์ง๊ณ ์๋์ง๋ฅผ ์ฒดํฌํฉ๋๋ค.
ONNX ๊ทธ๋ํ์ ์ ํจ์ฑ์ ๋ชจ๋ธ์ ๋ฒ์ , ๊ทธ๋ํ ๊ตฌ์กฐ, ๋
ธ๋๋ค, ๊ทธ๋ฆฌ๊ณ ์
๋ ฅ๊ฐ๊ณผ ์ถ๋ ฅ๊ฐ๋ค์ ๋ชจ๋ ์ฒดํฌํ์ฌ
๊ฒฐ์ ๋ฉ๋๋ค.</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="kn">import</span> <span class="nn">onnx</span>
<span class="n">onnx_model</span> <span class="o">=</span> <span class="n">onnx</span><span class="o">.</span><span class="n">load</span><span class="p">(</span><span class="s2">"super_resolution.onnx"</span><span class="p">)</span>
<span class="n">onnx</span><span class="o">.</span><span class="n">checker</span><span class="o">.</span><span class="n">check_model</span><span class="p">(</span><span class="n">onnx_model</span><span class="p">)</span>
</pre></div>
</div>
<p>์ด์ ONNX ๋ฐํ์์ Python API๋ฅผ ํตํด ๊ฒฐ๊ณผ๊ฐ์ ๊ณ์ฐํด๋ณด๋๋ก ํ๊ฒ ์ต๋๋ค.
์ด ๋ถ๋ถ์ ๋ณดํต ๋ณ๋์ ํ๋ก์ธ์ค ๋๋ ๋ณ๋์ ๋จธ์ ์์ ์คํ๋์ง๋ง, ์ด ํํ ๋ฆฌ์ผ์์๋ ๋ชจ๋ธ์ด
ONNX ๋ฐํ์๊ณผ PyTorch์์ ๋์ผํ ๊ฒฐ๊ณผ๋ฅผ ์ถ๋ ฅํ๋์ง๋ฅผ ํ์ธํ๊ธฐ ์ํด ๋์ผํ ํ๋ก์ธ์ค์์ ๊ณ์ ์คํํ๋๋ก
ํ๊ฒ ์ต๋๋ค.</p>
<p>๋ชจ๋ธ์ ONNX ๋ฐํ์์์ ์คํํ๊ธฐ ์ํด์๋ ๋ฏธ๋ฆฌ ์ค์ ๋ ์ธ์๋ค(๋ณธ ์์ ์์๋ ๊ธฐ๋ณธ๊ฐ์ ์ฌ์ฉํฉ๋๋ค)๋ก
๋ชจ๋ธ์ ์ํ ์ถ๋ก ์ธ์
์ ์์ฑํด์ผ ํฉ๋๋ค.
์ธ์
์ด ์์ฑ๋๋ฉด, ๋ชจ๋ธ์ run() API๋ฅผ ์ฌ์ฉํ์ฌ ๋ชจ๋ธ์ ์คํํฉ๋๋ค.
์ด API์ ๋ฆฌํด๊ฐ์ ONNX ๋ฐํ์์์ ์ฐ์ฐ๋ ๋ชจ๋ธ์ ๊ฒฐ๊ณผ๊ฐ๋ค์ ํฌํจํ๊ณ ์๋ ๋ฆฌ์คํธ์
๋๋ค.</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="kn">import</span> <span class="nn">onnxruntime</span>
<span class="n">ort_session</span> <span class="o">=</span> <span class="n">onnxruntime</span><span class="o">.</span><span class="n">InferenceSession</span><span class="p">(</span><span class="s2">"super_resolution.onnx"</span><span class="p">)</span>
<span class="k">def</span> <span class="nf">to_numpy</span><span class="p">(</span><span class="n">tensor</span><span class="p">):</span>
<span class="k">return</span> <span class="n">tensor</span><span class="o">.</span><span class="n">detach</span><span class="p">()</span><span class="o">.</span><span class="n">cpu</span><span class="p">()</span><span class="o">.</span><span class="n">numpy</span><span class="p">()</span> <span class="k">if</span> <span class="n">tensor</span><span class="o">.</span><span class="n">requires_grad</span> <span class="k">else</span> <span class="n">tensor</span><span class="o">.</span><span class="n">cpu</span><span class="p">()</span><span class="o">.</span><span class="n">numpy</span><span class="p">()</span>
<span class="c1"># ONNX ๋ฐํ์์์ ๊ณ์ฐ๋ ๊ฒฐ๊ณผ๊ฐ</span>
<span class="n">ort_inputs</span> <span class="o">=</span> <span class="p">{</span><span class="n">ort_session</span><span class="o">.</span><span class="n">get_inputs</span><span class="p">()[</span><span class="mi">0</span><span class="p">]</span><span class="o">.</span><span class="n">name</span><span class="p">:</span> <span class="n">to_numpy</span><span class="p">(</span><span class="n">x</span><span class="p">)}</span>
<span class="n">ort_outs</span> <span class="o">=</span> <span class="n">ort_session</span><span class="o">.</span><span class="n">run</span><span class="p">(</span><span class="kc">None</span><span class="p">,</span> <span class="n">ort_inputs</span><span class="p">)</span>
<span class="c1"># ONNX ๋ฐํ์๊ณผ PyTorch์์ ์ฐ์ฐ๋ ๊ฒฐ๊ณผ๊ฐ ๋น๊ต</span>
<span class="n">np</span><span class="o">.</span><span class="n">testing</span><span class="o">.</span><span class="n">assert_allclose</span><span class="p">(</span><span class="n">to_numpy</span><span class="p">(</span><span class="n">torch_out</span><span class="p">),</span> <span class="n">ort_outs</span><span class="p">[</span><span class="mi">0</span><span class="p">],</span> <span class="n">rtol</span><span class="o">=</span><span class="mf">1e-03</span><span class="p">,</span> <span class="n">atol</span><span class="o">=</span><span class="mf">1e-05</span><span class="p">)</span>
<span class="nb">print</span><span class="p">(</span><span class="s2">"Exported model has been tested with ONNXRuntime, and the result looks good!"</span><span class="p">)</span>
</pre></div>
</div>
<p>์ด์ PyTorch์ ONNX ๋ฐํ์์์ ์ฐ์ฐ๋ ๊ฒฐ๊ณผ๊ฐ์ด ์๋ก ์ผ์นํ๋์ง ์ค์ฐจ๋ฒ์ (rtol=1e-03, atol=1e-05)
์ด๋ด์์ ํ์ธํด์ผ ํฉ๋๋ค.
๋ง์ฝ ๊ฒฐ๊ณผ๊ฐ ์ผ์นํ์ง ์๋๋ค๋ฉด ONNX ๋ณํ๊ธฐ์ ๋ฌธ์ ๊ฐ ์๋ ๊ฒ์ด๋ ์ ํฌ์๊ฒ ์๋ ค์ฃผ์๊ธฐ ๋ฐ๋๋๋ค.</p>
<div class="section" id="onnx">
<h2>ONNX ๋ฐํ์์์ ์ด๋ฏธ์ง๋ฅผ ์
๋ ฅ๊ฐ์ผ๋ก ๋ชจ๋ธ์ ์คํํ๊ธฐ<a class="headerlink" href="#onnx" title="Permalink to this headline">ยถ</a></h2>
<p>์ง๊ธ๊น์ง PyTorch ๋ชจ๋ธ์ ๋ณํํ๊ณ ์ด๋ป๊ฒ ONNX ๋ฐํ์์์ ๊ตฌ๋ํ๋์ง ๊ฐ์์ ํ
์๋ฅผ ์
๋ ฅ๊ฐ์ผ๋ก ํ์ฌ
์ดํด๋ณด์์ต๋๋ค.</p>
<p>๋ณธ ํํ ๋ฆฌ์ผ์์๋ ์๋์ ๊ฐ์ ์ ๋ช
ํ ๊ณ ์์ด ์ฌ์ง์ ์ฌ์ฉํ๋๋ก ํ๊ฒ ์ต๋๋ค.</p>
<div class="figure">
<img alt="cat" src="../_images/cat_224x224.jpg" />
</div>
<p>๋จผ์ , PIL ๋ผ์ด๋ธ๋ฌ๋ฆฌ๋ฅผ ์ฌ์ฉํ์ฌ ์ด๋ฏธ์ง๋ฅผ ๋ก๋ํ๊ณ ์ ์ฒ๋ฆฌํ๊ฒ ์ต๋๋ค. ์ด ์ ์ฒ๋ฆฌ๋ ์ ๊ฒฝ๋ง ํ์ต๊ณผ ํ
์คํธ์
๋ณดํธ์ ์ผ๋ก ์ ์ฉ๋๊ณ ์๋ ์ ์ฒ๋ฆฌ ๊ณผ์ ์
๋๋ค.</p>
<dl class="docutils">
<dt>๋จผ์ ์ด๋ฏธ์ง๋ฅผ ๋ชจ๋ธ์ ์
๋ ฅ๊ฐ ํฌ๊ธฐ(224x224)์ ๋ง๊ฒ ๋ฆฌ์ฌ์ด์ฆํฉ๋๋ค.</dt>
<dd>๊ทธ๋ฆฌ๊ณ ์ด๋ฏธ์ง๋ฅผ Y, Cb, Cr ์ฑ๋ถ์ผ๋ก ๋ถํดํฉ๋๋ค.</dd>
</dl>
<p>Y ์ฑ๋ถ[์ญ์ ์ฃผ: ํ๋ ์ฑ๋ถ]์ ๊ทธ๋ ์ด์ค์ผ์ผ(ํ์์กฐ) ์ด๋ฏธ์ง๋ฅผ ๋ํ๋ด๊ณ , Cb ์ฑ๋ถ์ ํ๋์์์ ๋ฐ๊ธฐ๋ฅผ ๋บ ์์ฐจ ์ฑ๋ถ,
Cr์ ๋นจ๊ฐ์์์ ๋ฐ๊ธฐ๋ฅผ ๋บ ์์ฐจ ์ฑ๋ถ์ ๋ํ๋
๋๋ค.
์ฌ๋์ ๋์ Y ์ฑ๋ถ์ ๋ ๋ฏผ๊ฐํ๊ฒ ๋ฐ์ํ๊ธฐ ๋๋ฌธ์ ์ ํฌ์๊ฒ๋ ํ์ฌ ์ด ์ฑ๋ถ์ด ์ค์ํ๊ณ , ์ด Y ์ฑ๋ถ์ ๋ณํํ ๊ฒ์
๋๋ค.
Y ์ฑ๋ถ์ ๋ฝ์๋ธ ๋ค์, ์ถ์ถํ Y ์ฑ๋ถ์ ๋ชจ๋ธ์ ์
๋ ฅ๊ฐ์ด ๋ ํ
์๋ก ๋ณํํฉ๋๋ค.</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="kn">from</span> <span class="nn">PIL</span> <span class="kn">import</span> <span class="n">Image</span>
<span class="kn">import</span> <span class="nn">torchvision.transforms</span> <span class="k">as</span> <span class="nn">transforms</span>
<span class="n">img</span> <span class="o">=</span> <span class="n">Image</span><span class="o">.</span><span class="n">open</span><span class="p">(</span><span class="s2">"./_static/img/cat.jpg"</span><span class="p">)</span>
<span class="n">resize</span> <span class="o">=</span> <span class="n">transforms</span><span class="o">.</span><span class="n">Resize</span><span class="p">([</span><span class="mi">224</span><span class="p">,</span> <span class="mi">224</span><span class="p">])</span>
<span class="n">img</span> <span class="o">=</span> <span class="n">resize</span><span class="p">(</span><span class="n">img</span><span class="p">)</span>
<span class="n">img_ycbcr</span> <span class="o">=</span> <span class="n">img</span><span class="o">.</span><span class="n">convert</span><span class="p">(</span><span class="s1">'YCbCr'</span><span class="p">)</span>
<span class="n">img_y</span><span class="p">,</span> <span class="n">img_cb</span><span class="p">,</span> <span class="n">img_cr</span> <span class="o">=</span> <span class="n">img_ycbcr</span><span class="o">.</span><span class="n">split</span><span class="p">()</span>
<span class="n">to_tensor</span> <span class="o">=</span> <span class="n">transforms</span><span class="o">.</span><span class="n">ToTensor</span><span class="p">()</span>
<span class="n">img_y</span> <span class="o">=</span> <span class="n">to_tensor</span><span class="p">(</span><span class="n">img_y</span><span class="p">)</span>
<span class="n">img_y</span><span class="o">.</span><span class="n">unsqueeze_</span><span class="p">(</span><span class="mi">0</span><span class="p">)</span>
</pre></div>
</div>
<p>๋ค์์ ๋ฆฌ์ฌ์ด์ฆ๋ ๊ทธ๋ ์ด์ค์ผ์ผ ๊ณ ์์ด ์ด๋ฏธ์ง ํ
์๋ฅผ ์์ ์ค๋ช
ํ๋ ๊ฒ์ฒ๋ผ ์ดํด์๋ ๋ชจ๋ธ์ ๋๊ฒจ์ฃผ์ด
ONNX ๋ฐํ์์์ ์คํํ๋๋ก ํ๊ฒ ์ต๋๋ค.</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">ort_inputs</span> <span class="o">=</span> <span class="p">{</span><span class="n">ort_session</span><span class="o">.</span><span class="n">get_inputs</span><span class="p">()[</span><span class="mi">0</span><span class="p">]</span><span class="o">.</span><span class="n">name</span><span class="p">:</span> <span class="n">to_numpy</span><span class="p">(</span><span class="n">img_y</span><span class="p">)}</span>
<span class="n">ort_outs</span> <span class="o">=</span> <span class="n">ort_session</span><span class="o">.</span><span class="n">run</span><span class="p">(</span><span class="kc">None</span><span class="p">,</span> <span class="n">ort_inputs</span><span class="p">)</span>
<span class="n">img_out_y</span> <span class="o">=</span> <span class="n">ort_outs</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span>
</pre></div>
</div>
<p>์ด๋ ๊ฒ ํ๋ฉด ๋ชจ๋ธ์ ๊ฒฐ๊ณผ๊ฐ์ ํ
์๊ฐ ๋ฉ๋๋ค.
์ด์ ๋ชจ๋ธ์ ๊ฒฐ๊ณผ๊ฐ์ ์ฒ๋ฆฌํ์ฌ ๊ฒฐ๊ณผ๊ฐ ํ
์์์ ๋ง์ง๋ง ์ต์ข
์ถ๋ ฅ ์ด๋ฏธ์ง๋ฅผ ๋ง๋ค๊ณ ์ด๋ฅผ ์ ์ฅํด๋ณด๋๋ก ํ๊ฒ ์ต๋๋ค.
ํ์ฒ๋ฆฌ ๋จ๊ณ๋ <a class="reference external" href="https://github.com/pytorch/examples/blob/master/super_resolution/super_resolve.py">๋งํฌ</a> ์
๊ตฌํ๋์ด์๋ ์ดํด์๋ ๋ชจ๋ธ ์ฝ๋์์ ๊ฐ์ ธ์จ ๊ฒ์
๋๋ค.</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">img_out_y</span> <span class="o">=</span> <span class="n">Image</span><span class="o">.</span><span class="n">fromarray</span><span class="p">(</span><span class="n">np</span><span class="o">.</span><span class="n">uint8</span><span class="p">((</span><span class="n">img_out_y</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span> <span class="o">*</span> <span class="mf">255.0</span><span class="p">)</span><span class="o">.</span><span class="n">clip</span><span class="p">(</span><span class="mi">0</span><span class="p">,</span> <span class="mi">255</span><span class="p">)[</span><span class="mi">0</span><span class="p">]),</span> <span class="n">mode</span><span class="o">=</span><span class="s1">'L'</span><span class="p">)</span>
<span class="c1"># PyTorch ๋ฒ์ ์ ํ์ฒ๋ฆฌ ๊ณผ์ ์ฝ๋๋ฅผ ์ด์ฉํด ๊ฒฐ๊ณผ ์ด๋ฏธ์ง ๋ง๋ค๊ธฐ</span>
<span class="n">final_img</span> <span class="o">=</span> <span class="n">Image</span><span class="o">.</span><span class="n">merge</span><span class="p">(</span>
<span class="s2">"YCbCr"</span><span class="p">,</span> <span class="p">[</span>
<span class="n">img_out_y</span><span class="p">,</span>
<span class="n">img_cb</span><span class="o">.</span><span class="n">resize</span><span class="p">(</span><span class="n">img_out_y</span><span class="o">.</span><span class="n">size</span><span class="p">,</span> <span class="n">Image</span><span class="o">.</span><span class="n">BICUBIC</span><span class="p">),</span>
<span class="n">img_cr</span><span class="o">.</span><span class="n">resize</span><span class="p">(</span><span class="n">img_out_y</span><span class="o">.</span><span class="n">size</span><span class="p">,</span> <span class="n">Image</span><span class="o">.</span><span class="n">BICUBIC</span><span class="p">),</span>
<span class="p">])</span><span class="o">.</span><span class="n">convert</span><span class="p">(</span><span class="s2">"RGB"</span><span class="p">)</span>
<span class="c1"># ์ด๋ฏธ์ง๋ฅผ ์ ์ฅํ๊ณ ๋ชจ๋ฐ์ผ ๊ธฐ๊ธฐ์์์ ๊ฒฐ๊ณผ ์ด๋ฏธ์ง์ ๋น๊ตํ๊ธฐ</span>
<span class="n">final_img</span><span class="o">.</span><span class="n">save</span><span class="p">(</span><span class="s2">"./_static/img/cat_superres_with_ort.jpg"</span><span class="p">)</span>
</pre></div>
</div>
<div class="figure">
<img alt="output\_cat" src="../_images/cat_superres_with_ort.jpg" />
</div>
<p>ONNX ๋ฐํ์์ ํฌ๋ก์คํ๋ซํผ ์์ง์ผ๋ก์ CPU์ GPU ๋ฟ๋ง ์๋๋ผ ์ฌ๋ฌ ํ๋ซํผ์์ ์คํ๋ ์ ์์ต๋๋ค.</p>
<p>ONNX ๋ฐํ์์ Azure Machine Learning Services์ ๊ฐ์ ํด๋ผ์ฐ๋์ ๋ฐฐํฌ๋์ด ๋ชจ๋ธ ์ถ๋ก ์ ํ๋๋ฐ
์ฌ์ฉ๋ ์๋ ์์ต๋๋ค.
๋ ์์ธํ ๋ด์ฉ์ <a class="reference external" href="https://docs.microsoft.com/en-us/azure/machine-learning/service/concept-onnx">๋งํฌ</a> ๋ฅผ
์ฐธ์กฐํด์ฃผ์ญ์์ค.</p>
<p>ONNX ๋ฐํ์์ ์ฑ๋ฅ์ ๊ดํ ๊ฒ์ <a class="reference external" href="https://github.com/microsoft/onnxruntime#high-performance">์ฌ๊ธฐ</a> ์์
ํ์ธํ์ค ์ ์์ต๋๋ค.</p>
<p>ONNX ๋ฐํ์์ ๊ดํ ๋ ์์ธํ ๋ด์ฉ์ <a class="reference external" href="https://github.com/microsoft/onnxruntime">์ด ๊ณณ</a> ์ ์ฐธ์กฐํด ์ฃผ์ธ์.</p>
<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 0 minutes 0.000 seconds)</p>
<div class="sphx-glr-footer class sphx-glr-footer-example docutils container" id="sphx-glr-download-advanced-super-resolution-with-onnxruntime-py">
<div class="sphx-glr-download docutils container">
<a class="reference download internal" download="" href="../_downloads/f6608f362ad07183c0ee305ce6aaa917/super_resolution_with_onnxruntime.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">super_resolution_with_onnxruntime.py</span></code></a></div>
<div class="sphx-glr-download docutils container">
<a class="reference download internal" download="" href="../_downloads/dba562a36c81e67ddf3b2f9503b7009b/super_resolution_with_onnxruntime.ipynb"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Jupyter</span> <span class="pre">notebook:</span> <span class="pre">super_resolution_with_onnxruntime.ipynb</span></code></a></div>
</div>
<p class="sphx-glr-signature"><a class="reference external" href="https://sphinx-gallery.readthedocs.io">Gallery generated by Sphinx-Gallery</a></p>
</div>
</div>
</article>
</div>
<footer>
<div class="rst-footer-buttons" role="navigation" aria-label="footer navigation">
<a href="../intermediate/fx_conv_bn_fuser.html" class="btn btn-neutral float-right" title="(๋ฒ ํ) FX์์ ํฉ์ฑ๊ณฑ/๋ฐฐ์น ์ ๊ทํ(Convolution/Batch Norm) ๊ฒฐํฉ๊ธฐ(Fuser) ๋ง๋ค๊ธฐ" accesskey="n" rel="next">Next <img src="../_static/images/chevron-right-orange.svg" class="next-page"></a>
<a href="cpp_export.html" class="btn btn-neutral" title="C++์์ TorchScript ๋ชจ๋ธ ๋ก๋ฉํ๊ธฐ" accesskey="p" rel="prev"><img src="../_static/images/chevron-right-orange.svg" class="previous-page"> Previous</a>
</div>
<hr class="rating-hr hr-top">
<div class="rating-container">
<div class="rating-prompt">Rate this Tutorial</div>
<div class="stars-outer">
<i class="far fa-star" title="1 Star" data-behavior="tutorial-rating" data-count="1"></i>
<i class="far fa-star" title="2 Stars" data-behavior="tutorial-rating" data-count="2"></i>
<i class="far fa-star" title="3 Stars" data-behavior="tutorial-rating" data-count="3"></i>
<i class="far fa-star" title="4 Stars" data-behavior="tutorial-rating" data-count="4"></i>
<i class="far fa-star" title="5 Stars" data-behavior="tutorial-rating" data-count="5"></i>
</div>
</div>
<hr class="rating-hr hr-bottom"/>
<div role="contentinfo">
<p>
© Copyright 2021, PyTorch & PyTorch Korea Community.
</p>
</div>
<div>
Built with <a href="http://sphinx-doc.org/">Sphinx</a> using a <a href="https://github.com/rtfd/sphinx_rtd_theme">theme</a> provided by <a href="https://readthedocs.org">Read the Docs</a>.
</div>
</footer>
</div>
</div>
<div class="pytorch-content-right" id="pytorch-content-right">
<div class="pytorch-right-menu" id="pytorch-right-menu">
<div class="pytorch-side-scroll" id="pytorch-side-scroll-right">
<ul>
<li><a class="reference internal" href="#">(์ ํ) PyTorch ๋ชจ๋ธ์ ONNX์ผ๋ก ๋ณํํ๊ณ ONNX ๋ฐํ์์์ ์คํํ๊ธฐ</a><ul>
<li><a class="reference internal" href="#onnx">ONNX ๋ฐํ์์์ ์ด๋ฏธ์ง๋ฅผ ์
๋ ฅ๊ฐ์ผ๋ก ๋ชจ๋ธ์ ์คํํ๊ธฐ</a></li>
</ul>
</li>
</ul>
</div>
</div>
</div>
</section>
</div>
<script type="text/javascript" id="documentation_options" data-url_root="../" src="../_static/documentation_options.js"></script>
<script type="text/javascript" src="../_static/jquery.js"></script>
<script type="text/javascript" src="../_static/underscore.js"></script>
<script type="text/javascript" src="../_static/doctools.js"></script>
<script type="text/javascript" src="../_static/clipboard.min.js"></script>
<script type="text/javascript" src="../_static/copybutton.js"></script>
<script type="text/javascript" src="https://cdn.jsdelivr.net/npm/katex@0.13.11/dist/katex.min.js"></script>
<script type="text/javascript" src="https://cdn.jsdelivr.net/npm/katex@0.13.11/dist/contrib/auto-render.min.js"></script>
<script type="text/javascript" src="../_static/katex_autorenderer.js"></script>
<script type="text/javascript" src="../_static/js/vendor/popper.min.js"></script>
<script type="text/javascript" src="../_static/js/vendor/bootstrap.min.js"></script>
<script src="https://cdnjs.cloudflare.com/ajax/libs/list.js/1.5.0/list.min.js"></script>
<script type="text/javascript" src="../_static/js/theme.js"></script>
<script type="text/javascript">
jQuery(function () {
SphinxRtdTheme.Navigation.enable(true);
});
</script>
<script>
//add microsoft link
if(window.location.href.indexOf("/beginner/basics/")!= -1)
{
var url="https://docs.microsoft.com/learn/paths/pytorch-fundamentals/?wt.mc_id=aiml-7486-cxa";
switch(window.location.pathname.split("/").pop().replace('.html',''))
{
case"quickstart_tutorial":
url="https://docs.microsoft.com/learn/modules/intro-machine-learning-pytorch/9-quickstart?WT.mc_id=aiml-7486-cxa";
break;
case"tensorqs_tutorial":
url="https://docs.microsoft.com/learn/modules/intro-machine-learning-pytorch/2-tensors?WT.mc_id=aiml-7486-cxa";
break;
case"data_tutorial":
url="https://docs.microsoft.com/learn/modules/intro-machine-learning-pytorch/3-data?WT.mc_id=aiml-7486-cxa";
break;
case"transforms_tutorial":
url="https://docs.microsoft.com/learn/modules/intro-machine-learning-pytorch/4-transforms?WT.mc_id=aiml-7486-cxa";
break;
case"buildmodel_tutorial":
url="https://docs.microsoft.com/learn/modules/intro-machine-learning-pytorch/5-model?WT.mc_id=aiml-7486-cxa";
break;
case"autogradqs_tutorial":
url="https://docs.microsoft.com/learn/modules/intro-machine-learning-pytorch/6-autograd?WT.mc_id=aiml-7486-cxa";
break;
case"optimization_tutorial":
url="https://docs.microsoft.com/learn/modules/intro-machine-learning-pytorch/7-optimization?WT.mc_id=aiml-7486-cxa";
break;
case"saveloadrun_tutorial":
url="https://docs.microsoft.com/learn/modules/intro-machine-learning-pytorch/8-inference?WT.mc_id=aiml-7486-cxa";
}
$(".pytorch-call-to-action-links").children().first().before("<a href="+url+' data-behavior="call-to-action-event" data-response="Run in Microsoft Learn" target="_blank"><div id="microsoft-learn-link" style="padding-bottom: 0.625rem;border-bottom: 1px solid #f3f4f7;padding-right: 2.5rem;display: -webkit-box; display: -ms-flexbox; isplay: flex; -webkit-box-align: center;-ms-flex-align: center;align-items: center;"><img class="call-to-action-img" src="../../_static/images/microsoft-logo.svg"/><div class="call-to-action-desktop-view">Run in Microsoft Learn</div><div class="call-to-action-mobile-view">Learn</div></div></a>')
}
</script>
<script async src="https://www.googletagmanager.com/gtag/js?id=UA-71919972-3"></script>
<script>
window.dataLayer = window.dataLayer || [];
function gtag(){dataLayer.push(arguments);}
gtag('js', new Date());
gtag('config', 'UA-71919972-3');
</script>
<script>
$("[data-behavior='call-to-action-event']").on('click', function(){
ga('send', {
hitType: 'event',
eventCategory: $(this).attr("data-response"),
eventAction: 'click',
eventLabel: window.location.href
});
gtag('event', 'click', {
'event_category': $(this).attr("data-response"),
'event_label': $("h1").first().text(),
'tutorial_link': window.location.href
});
});
$("[data-behavior='tutorial-rating']").on('click', function(){
gtag('event', 'click', {
'event_category': 'Tutorial Rating',
'event_label': $("h1").first().text(),
'value': $(this).attr("data-count")
});
});
if (location.pathname == "/") {
$(".rating-container").hide();
$(".hr-bottom").hide();
}
</script>
<script type="text/javascript">
var collapsedSections = ['ํ์ดํ ์น(PyTorch) ๋ ์ํผ', 'ํ์ดํ ์น(PyTorch) ๋ฐฐ์ฐ๊ธฐ', '์ด๋ฏธ์ง/๋น๋์ค', '์ค๋์ค', 'ํ
์คํธ', '๊ฐํํ์ต', 'PyTorch ๋ชจ๋ธ์ ํ๋ก๋์
ํ๊ฒฝ์ ๋ฐฐํฌํ๊ธฐ', 'Code Transforms with FX', 'ํ๋ก ํธ์๋ API', 'PyTorch ํ์ฅํ๊ธฐ', '๋ชจ๋ธ ์ต์ ํ', '๋ณ๋ ฌ ๋ฐ ๋ถ์ฐ ํ์ต', 'Mobile'];
</script>
<!-- Begin Footer -->
<div class="container-fluid docs-tutorials-resources" id="docs-tutorials-resources">
<div class="container">
<div class="row">
<div class="col-md-4 text-center">
<h2>๊ณต์ ๋ฌธ์ (์์ด)</h2>
<p>PyTorch ๊ณต์ ๋ฌธ์์
๋๋ค.</p>
<a id="orgTutorialLink" class="with-right-arrow" href="https://pytorch.org/docs/stable/index.html" target="_blank">๊ณต์ ๋ฌธ์๋ก ์ด๋</a>
</div>
<div class="col-md-4 text-center">
<h2>ํ๊ตญ์ด ํํ ๋ฆฌ์ผ</h2>
<p>ํ๊ตญ์ด๋ก ๋ฒ์ญ ์ค์ธ PyTorch ํํ ๋ฆฌ์ผ์
๋๋ค.</p>
<a class="with-right-arrow" href="https://tutorials.pytorch.kr">ํํ ๋ฆฌ์ผ๋ก ์ด๋</a>
</div>
<div class="col-md-4 text-center">
<h2>์ปค๋ฎค๋ํฐ</h2>
<p>๋ค๋ฅธ ์ฌ์ฉ์๋ค๊ณผ ์๊ฒฌ์ ๋๋ ๋ณด์ธ์!</p>
<a class="with-right-arrow" href="https://discuss.pytorch.kr">์ปค๋ฎค๋ํฐ๋ก ์ด๋</a>
</div>
</div>
</div>
</div>
<footer class="site-footer">
<div class="container footer-container">
<div class="footer-logo-wrapper">
<a href="https://pytorch.kr/" class="footer-logo"></a>
</div>
<div class="footer-links-wrapper">
<div class="footer-links-col">
<ul>
<li class="list-title"><a href="https://pytorch.org">PyTorch ํํ์ด์ง (๊ณต์)</a></li>
<li><a href="https://pytorch.org" target="_blank">๊ณต์ ํํ์ด์ง</a></li>
<li><a href="https://pytorch.org/tutorials" target="_blank">๊ณต์ ํํ ๋ฆฌ์ผ</a></li>
<li><a href="https://pytorch.org/docs" target="_blank">๊ณต์ ๋ฌธ์</a></li>
</ul>
</div>
<div class="footer-links-col">
<ul>
<li class="list-title"><a href="">ํ๊ตญ ์ฌ์ฉ์ ๋ชจ์</a></li>
<li><a href="https://pytorch.kr/about">์ฌ์ดํธ ์๊ฐ</a></li>
<li><a href="https://tutorials.pytorch.kr/">ํ๊ตญ์ด ํํ ๋ฆฌ์ผ</a></li>
<li><a href="https://github.com/9bow/PyTorch-tutorials-kr" target="_blank">ํ๊ตญ์ด ํํ ๋ฆฌ์ผ ์ ์ฅ์</a></li>
</ul>
</div>
</div>
<div class="trademark-disclaimer">
<ul>
<li>์ด ์ฌ์ดํธ๋ PyTorch ํ๊ตญ ์ฌ์ฉ์ ์ปค๋ฎค๋ํฐ๋ก Facebook, Inc์์ ์ด์ํ๋ ์ฌ์ดํธ๊ฐ ์๋๋๋ค. PyTorch, PyTorch ๋ก๊ณ ๋ฐ ๋ชจ๋ ๊ด๋ จ ํ๊ธฐ๋ Facebook, Inc์ ์ํ์
๋๋ค.</li>
<li>This site is a user community and is not operated by Facebook, Inc. PyTorch, the PyTorch logo and any related marks are trademarks of Facebook, Inc.</li>
</ul>
</div>
</div>
</footer>
<!-- End Footer -->
<!-- Begin Mobile Menu -->
<div class="mobile-main-menu">
<div class="container-fluid">
<div class="container">
<div class="mobile-main-menu-header-container">
<a class="header-logo" href="https://pytorch.kr/" aria-label="PyTorch"></a>
<a class="main-menu-close-button" href="#" data-behavior="close-mobile-menu"></a>
</div>
</div>
</div>
<div class="mobile-main-menu-links-container">
<div class="main-menu">
<ul>
<li>
<a href="https://pytorch.kr/get-started">์์ํ๊ธฐ</a>
</li>
<li class="active">
<a href="https://tutorials.pytorch.kr">ํํ ๋ฆฌ์ผ</a>
</li>
<li>
<a href="https://pytorch.kr/hub">ํ๋ธ</a>
</li>
<li>
<a href="https://discuss.pytorch.kr">์ปค๋ฎค๋ํฐ</a>
</li>
</ul>
</div>
</div>
</div>
<!-- End Mobile Menu -->
<script type="text/javascript" src="../_static/js/vendor/anchor.min.js"></script>
<script type="text/javascript">
$(document).ready(function() {
mobileMenu.bind();
mobileTOC.bind();
pytorchAnchors.bind();
sideMenus.bind();
scrollToAnchor.bind();
highlightNavigation.bind();
mainMenuDropdown.bind();
filterTags.bind();
// Add class to links that have code blocks, since we cannot create links in code blocks
$("article.pytorch-article a span.pre").each(function(e) {
$(this).closest("a").addClass("has-code");
});
})
</script>
</body>
</html>