-
Notifications
You must be signed in to change notification settings - Fork 1
/
17129387431530.html
801 lines (630 loc) · 28.3 KB
/
17129387431530.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
<!doctype html>
<html class="no-js" lang="en">
<head>
<meta charset="utf-8" />
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
<title>
场景文本识别算法-NRTR -
</title>
<link href="atom.xml" rel="alternate" title="" type="application/atom+xml">
<link rel="stylesheet" href="asset/css/foundation.min.css" />
<link rel="stylesheet" href="asset/css/docs.css" />
<script src="asset/js/vendor/modernizr.js"></script>
<script src="asset/js/vendor/jquery.js"></script>
<script type="text/javascript">
function before_search(){
var searchVal = 'site: ' + document.getElementById('search_input').value;
document.getElementById('search_q').value = searchVal;
return true;
}
</script>
</head>
<body class="antialiased hide-extras">
<div class="marketing off-canvas-wrap" data-offcanvas>
<div class="inner-wrap">
<nav class="top-bar docs-bar hide-for-small" data-topbar>
<section class="top-bar-section">
<div class="row">
<div style="position: relative;width:100%;"><div style="position: absolute; width:100%;">
<ul id="main-menu" class="left">
<li id=""><a target="_self" href="index.html">Home</a></li>
<li id=""><a target="_self" href="archives.html">Archives</a></li>
</ul>
<ul class="right" id="search-wrap">
<li>
<form target="_blank" onsubmit="return before_search();" action="https://google.com/search" method="get">
<input type="hidden" id="search_q" name="q" value="" />
<input tabindex="1" type="search" id="search_input" placeholder="Search"/>
</form>
</li>
</ul>
</div></div>
</div>
</section>
</nav>
<nav class="tab-bar show-for-small">
<a href="javascript:void(0)" class="left-off-canvas-toggle menu-icon">
<span> </span>
</a>
</nav>
<aside class="left-off-canvas-menu">
<ul class="off-canvas-list">
<li><a target="_self" href="index.html">Home</a></li>
<li><a target="_self" href="archives.html">Archives</a></li>
<li><label>Categories</label></li>
</ul>
</aside>
<a class="exit-off-canvas" href="#"></a>
<section id="main-content" role="main" class="scroll-container">
<script type="text/javascript">
$(function(){
$('#menu_item_index').addClass('is_active');
});
</script>
<div class="row">
<div class="large-8 medium-8 columns">
<div class="markdown-body article-wrap">
<div class="article">
<h1>场景文本识别算法-NRTR</h1>
<div class="read-more clearfix">
<span class="date">2024/04/13</span>
<span class="comments">
</span>
</div>
</div><!-- article -->
<div class="article-content">
<ul>
<li><a href="#1">1. 算法简介</a></li>
<li><a href="#2">2. 环境配置</a></li>
<li><a href="#3">3. 模型训练、评估、预测</a>
<ul>
<li><a href="#3-1">3.1 训练</a></li>
<li><a href="#3-2">3.2 评估</a></li>
<li><a href="#3-3">3.3 预测</a></li>
</ul>
</li>
<li><a href="#4">4. 推理部署</a>
<ul>
<li><a href="#4-1">4.1 Python推理</a></li>
<li><a href="#4-2">4.2 C++推理</a></li>
<li><a href="#4-3">4.3 Serving服务化部署</a></li>
<li><a href="#4-4">4.4 更多推理部署</a></li>
</ul>
</li>
<li><a href="#5">5. FAQ</a></li>
<li><a href="#6">6. 发行公告</a></li>
</ul>
<p><a name="1"></a></p>
<h2><a id="1%E7%AE%97%E6%B3%95%E7%AE%80%E4%BB%8B" class="anchor" aria-hidden="true"><span class="octicon octicon-link"></span></a>1. 算法简介</h2>
<p>论文信息：</p>
<blockquote>
<p><a href="https://arxiv.org/abs/1806.00926">NRTR: A No-Recurrence Sequence-to-Sequence Model For Scene Text Recognition</a><br />
Fenfen Sheng and Zhineng Chen and Bo Xu<br />
ICDAR, 2019</p>
</blockquote>
<p><a name="model"></a><br />
<code>NRTR</code>使用MJSynth和SynthText两个文字识别数据集训练，在IIIT, SVT, IC03, IC13, IC15, SVTP, CUTE数据集上进行评估，算法复现效果如下：</p>
<table>
<thead>
<tr>
<th>模型</th>
<th>骨干网络</th>
<th>配置文件</th>
<th>Acc</th>
<th>下载链接</th>
</tr>
</thead>
<tbody>
<tr>
<td>NRTR</td>
<td>MTB</td>
<td><a href="../../configs/rec/rec_mtb_nrtr.yml">rec_mtb_nrtr.yml</a></td>
<td>84.21%</td>
<td><a href="https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_mtb_nrtr_train.tar">训练模型</a></td>
</tr>
</tbody>
</table>
<p><a name="2"></a></p>
<h2><a id="2%E7%8E%AF%E5%A2%83%E9%85%8D%E7%BD%AE" class="anchor" aria-hidden="true"><span class="octicon octicon-link"></span></a>2. 环境配置</h2>
<p>请先参考<a href="media/17117697540963/environment.md">《运行环境准备》</a>配置PaddleOCR运行环境，参考<a href="media/17117697540963/clone.md">《项目克隆》</a>克隆项目代码。</p>
<p><a name="3"></a></p>
<h2><a id="3%E6%A8%A1%E5%9E%8B%E8%AE%AD%E7%BB%83%E3%80%81%E8%AF%84%E4%BC%B0%E3%80%81%E9%A2%84%E6%B5%8B" class="anchor" aria-hidden="true"><span class="octicon octicon-link"></span></a>3. 模型训练、评估、预测</h2>
<p><a name="3-1"></a></p>
<h3><a id="3-1%E6%A8%A1%E5%9E%8B%E8%AE%AD%E7%BB%83" class="anchor" aria-hidden="true"><span class="octicon octicon-link"></span></a>3.1 模型训练</h3>
<p>请参考<a href="media/17117697540963/recognition.md">文本识别训练教程</a>。PaddleOCR对代码进行了模块化，训练<code>NRTR</code>识别模型时需要<strong>更换配置文件</strong>为<code>NRTR</code>的<a href="../../configs/rec/rec_mtb_nrtr.yml">配置文件</a>。</p>
<h4><a id="%E5%90%AF%E5%8A%A8%E8%AE%AD%E7%BB%83" class="anchor" aria-hidden="true"><span class="octicon octicon-link"></span></a>启动训练</h4>
<p>具体地，在完成数据准备后，便可以启动训练，训练命令如下：</p>
<pre class="line-numbers"><code class="language-shell">#单卡训练（训练周期长，不建议）
python3 tools/train.py -c configs/rec/rec_mtb_nrtr.yml
#多卡训练，通过--gpus参数指定卡号
python3 -m paddle.distributed.launch --gpus '0,1,2,3' tools/train.py -c configs/rec/rec_mtb_nrtr.yml
</code></pre>
<p><a name="3-2"></a></p>
<h3><a id="3-2%E8%AF%84%E4%BC%B0" class="anchor" aria-hidden="true"><span class="octicon octicon-link"></span></a>3.2 评估</h3>
<p>可下载已训练完成的<a href="#model">模型文件</a>，使用如下命令进行评估：</p>
<pre class="line-numbers"><code class="language-shell"># 注意将pretrained_model的路径设置为本地路径。
python3 -m paddle.distributed.launch --gpus '0' tools/eval.py -c configs/rec/rec_mtb_nrtr.yml -o Global.pretrained_model=./rec_mtb_nrtr_train/best_accuracy
</code></pre>
<p><a name="3-3"></a></p>
<h3><a id="3-3%E9%A2%84%E6%B5%8B" class="anchor" aria-hidden="true"><span class="octicon octicon-link"></span></a>3.3 预测</h3>
<p>使用如下命令进行单张图片预测：</p>
<pre class="line-numbers"><code class="language-shell"># 注意将pretrained_model的路径设置为本地路径。
python3 tools/infer_rec.py -c configs/rec/rec_mtb_nrtr.yml -o Global.infer_img='./doc/imgs_words_en/word_10.png' Global.pretrained_model=./rec_mtb_nrtr_train/best_accuracy
# 预测文件夹下所有图像时，可修改infer_img为文件夹，如 Global.infer_img='./doc/imgs_words_en/'。
</code></pre>
<p><a name="4"></a></p>
<h2><a id="4%E6%8E%A8%E7%90%86%E9%83%A8%E7%BD%B2" class="anchor" aria-hidden="true"><span class="octicon octicon-link"></span></a>4. 推理部署</h2>
<p><a name="4-1"></a></p>
<h3><a id="4-1-python%E6%8E%A8%E7%90%86" class="anchor" aria-hidden="true"><span class="octicon octicon-link"></span></a>4.1 Python推理</h3>
<p>首先将训练得到best模型，转换成inference model。这里以训练完成的模型为例（<a href="https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_mtb_nrtr_train.tar">模型下载地址</a> )，可以使用如下命令进行转换：</p>
<pre class="line-numbers"><code class="language-shell"># 注意将pretrained_model的路径设置为本地路径。
python3 tools/export_model.py -c configs/rec/rec_mtb_nrtr.yml -o Global.pretrained_model=./rec_mtb_nrtr_train/best_accuracy Global.save_inference_dir=./inference/rec_mtb_nrtr/
</code></pre>
<p><strong>注意：</strong></p>
<ul>
<li>如果您是在自己的数据集上训练的模型，并且调整了字典文件，请注意修改配置文件中的<code>character_dict_path</code>是否是所需要的字典文件。</li>
<li>如果您修改了训练时的输入大小，请修改<code>tools/export_model.py</code>文件中的对应NRTR的<code>infer_shape</code>。</li>
</ul>
<p>转换成功后，在目录下有三个文件：</p>
<pre class="line-numbers"><code class="language-plain_text">/inference/rec_mtb_nrtr/
├── inference.pdiparams # 识别inference模型的参数文件
├── inference.pdiparams.info # 识别inference模型的参数信息，可忽略
└── inference.pdmodel # 识别inference模型的program文件
</code></pre>
<p>执行如下命令进行模型推理：</p>
<pre class="line-numbers"><code class="language-shell">python3 tools/infer/predict_rec.py --image_dir='./doc/imgs_words_en/word_10.png' --rec_model_dir='./inference/rec_mtb_nrtr/' --rec_algorithm='NRTR' --rec_image_shape='1,32,100' --rec_char_dict_path='./ppocr/utils/EN_symbol_dict.txt'
# 预测文件夹下所有图像时，可修改image_dir为文件夹，如 --image_dir='./doc/imgs_words_en/'。
</code></pre>
<p><img src="../imgs_words_en/word_10.png" alt="" /></p>
<p>执行命令后，上面图像的预测结果（识别的文本和得分）会打印到屏幕上，示例如下：<br />
结果如下：</p>
<pre class="line-numbers"><code class="language-shell">Predicts of ./doc/imgs_words_en/word_10.png:('pain', 0.9465042352676392)
</code></pre>
<p><strong>注意</strong>：</p>
<ul>
<li>训练上述模型采用的图像分辨率是[1，32，100]，需要通过参数<code>rec_image_shape</code>设置为您训练时的识别图像形状。</li>
<li>在推理时需要设置参数<code>rec_char_dict_path</code>指定字典，如果您修改了字典，请修改该参数为您的字典文件。</li>
<li>如果您修改了预处理方法，需修改<code>tools/infer/predict_rec.py</code>中NRTR的预处理为您的预处理方法。</li>
</ul>
<p><a name="4-2"></a></p>
<h3><a id="4-2-c%E6%8E%A8%E7%90%86%E9%83%A8%E7%BD%B2" class="anchor" aria-hidden="true"><span class="octicon octicon-link"></span></a>4.2 C++推理部署</h3>
<p>由于C++预处理后处理还未支持NRTR，所以暂未支持</p>
<p><a name="4-3"></a></p>
<h3><a id="4-3-serving%E6%9C%8D%E5%8A%A1%E5%8C%96%E9%83%A8%E7%BD%B2" class="anchor" aria-hidden="true"><span class="octicon octicon-link"></span></a>4.3 Serving服务化部署</h3>
<p>暂不支持</p>
<p><a name="4-4"></a></p>
<h3><a id="4-4%E6%9B%B4%E5%A4%9A%E6%8E%A8%E7%90%86%E9%83%A8%E7%BD%B2" class="anchor" aria-hidden="true"><span class="octicon octicon-link"></span></a>4.4 更多推理部署</h3>
<p>暂不支持</p>
<p><a name="5"></a></p>
<h2><a id="5-faq" class="anchor" aria-hidden="true"><span class="octicon octicon-link"></span></a>5. FAQ</h2>
<ol>
<li><code>NRTR</code>论文中使用Beam搜索进行解码字符，但是速度较慢，这里默认未使用Beam搜索，以贪婪搜索进行解码字符。</li>
</ol>
<p><a name="6"></a></p>
<h2><a id="6%E5%8F%91%E8%A1%8C%E5%85%AC%E5%91%8A" class="anchor" aria-hidden="true"><span class="octicon octicon-link"></span></a>6. 发行公告</h2>
<ol>
<li>release/2.6更新NRTR代码结构，新版NRTR可加载旧版(release/2.5及之前)模型参数，使用下面示例代码将旧版模型参数转换为新版模型参数：</li>
</ol>
<pre class="line-numbers"><code class="language-python">
params = paddle.load('path/' + '.pdparams') # 旧版本参数
state_dict = model.state_dict() # 新版模型参数
new_state_dict = {}
for k1, v1 in state_dict.items():
k = k1
if 'encoder' in k and 'self_attn' in k and 'qkv' in k and 'weight' in k:
k_para = k[:13] + 'layers.' + k[13:]
q = params[k_para.replace('qkv', 'conv1')].transpose((1, 0, 2, 3))
k = params[k_para.replace('qkv', 'conv2')].transpose((1, 0, 2, 3))
v = params[k_para.replace('qkv', 'conv3')].transpose((1, 0, 2, 3))
new_state_dict[k1] = np.concatenate([q[:, :, 0, 0], k[:, :, 0, 0], v[:, :, 0, 0]], -1)
elif 'encoder' in k and 'self_attn' in k and 'qkv' in k and 'bias' in k:
k_para = k[:13] + 'layers.' + k[13:]
q = params[k_para.replace('qkv', 'conv1')]
k = params[k_para.replace('qkv', 'conv2')]
v = params[k_para.replace('qkv', 'conv3')]
new_state_dict[k1] = np.concatenate([q, k, v], -1)
elif 'encoder' in k and 'self_attn' in k and 'out_proj' in k:
k_para = k[:13] + 'layers.' + k[13:]
new_state_dict[k1] = params[k_para]
elif 'encoder' in k and 'norm3' in k:
k_para = k[:13] + 'layers.' + k[13:]
new_state_dict[k1] = params[k_para.replace('norm3', 'norm2')]
elif 'encoder' in k and 'norm1' in k:
k_para = k[:13] + 'layers.' + k[13:]
new_state_dict[k1] = params[k_para]
elif 'decoder' in k and 'self_attn' in k and 'qkv' in k and 'weight' in k:
k_para = k[:13] + 'layers.' + k[13:]
q = params[k_para.replace('qkv', 'conv1')].transpose((1, 0, 2, 3))
k = params[k_para.replace('qkv', 'conv2')].transpose((1, 0, 2, 3))
v = params[k_para.replace('qkv', 'conv3')].transpose((1, 0, 2, 3))
new_state_dict[k1] = np.concatenate([q[:, :, 0, 0], k[:, :, 0, 0], v[:, :, 0, 0]], -1)
elif 'decoder' in k and 'self_attn' in k and 'qkv' in k and 'bias' in k:
k_para = k[:13] + 'layers.' + k[13:]
q = params[k_para.replace('qkv', 'conv1')]
k = params[k_para.replace('qkv', 'conv2')]
v = params[k_para.replace('qkv', 'conv3')]
new_state_dict[k1] = np.concatenate([q, k, v], -1)
elif 'decoder' in k and 'self_attn' in k and 'out_proj' in k:
k_para = k[:13] + 'layers.' + k[13:]
new_state_dict[k1] = params[k_para]
elif 'decoder' in k and 'cross_attn' in k and 'q' in k and 'weight' in k:
k_para = k[:13] + 'layers.' + k[13:]
k_para = k_para.replace('cross_attn', 'multihead_attn')
q = params[k_para.replace('q', 'conv1')].transpose((1, 0, 2, 3))
new_state_dict[k1] = q[:, :, 0, 0]
elif 'decoder' in k and 'cross_attn' in k and 'q' in k and 'bias' in k:
k_para = k[:13] + 'layers.' + k[13:]
k_para = k_para.replace('cross_attn', 'multihead_attn')
q = params[k_para.replace('q', 'conv1')]
new_state_dict[k1] = q
elif 'decoder' in k and 'cross_attn' in k and 'kv' in k and 'weight' in k:
k_para = k[:13] + 'layers.' + k[13:]
k_para = k_para.replace('cross_attn', 'multihead_attn')
k = params[k_para.replace('kv', 'conv2')].transpose((1, 0, 2, 3))
v = params[k_para.replace('kv', 'conv3')].transpose((1, 0, 2, 3))
new_state_dict[k1] = np.concatenate([k[:, :, 0, 0], v[:, :, 0, 0]], -1)
elif 'decoder' in k and 'cross_attn' in k and 'kv' in k and 'bias' in k:
k_para = k[:13] + 'layers.' + k[13:]
k_para = k_para.replace('cross_attn', 'multihead_attn')
k = params[k_para.replace('kv', 'conv2')]
v = params[k_para.replace('kv', 'conv3')]
new_state_dict[k1] = np.concatenate([k, v], -1)
elif 'decoder' in k and 'cross_attn' in k and 'out_proj' in k:
k_para = k[:13] + 'layers.' + k[13:]
k_para = k_para.replace('cross_attn', 'multihead_attn')
new_state_dict[k1] = params[k_para]
elif 'decoder' in k and 'norm' in k:
k_para = k[:13] + 'layers.' + k[13:]
new_state_dict[k1] = params[k_para]
elif 'mlp' in k and 'weight' in k:
k_para = k[:13] + 'layers.' + k[13:]
k_para = k_para.replace('fc', 'conv')
k_para = k_para.replace('mlp.', '')
w = params[k_para].transpose((1, 0, 2, 3))
new_state_dict[k1] = w[:, :, 0, 0]
elif 'mlp' in k and 'bias' in k:
k_para = k[:13] + 'layers.' + k[13:]
k_para = k_para.replace('fc', 'conv')
k_para = k_para.replace('mlp.', '')
w = params[k_para]
new_state_dict[k1] = w
else:
new_state_dict[k1] = params[k1]
if list(new_state_dict[k1].shape) != list(v1.shape):
print(k1)
for k, v1 in state_dict.items():
if k not in new_state_dict.keys():
print(1, k)
elif list(new_state_dict[k].shape) != list(v1.shape):
print(2, k)
model.set_state_dict(new_state_dict)
paddle.save(model.state_dict(), 'nrtrnew_from_old_params.pdparams')
</code></pre>
<ol start="2">
<li>新版相比与旧版，代码结构简洁，推理速度有所提高。</li>
</ol>
<h2><a id="%E5%BC%95%E7%94%A8" class="anchor" aria-hidden="true"><span class="octicon octicon-link"></span></a>引用</h2>
<pre class="line-numbers"><code class="language-bibtex">@article{Sheng2019NRTR,
title = {NRTR: A No-Recurrence Sequence-to-Sequence Model For Scene Text Recognition},
author = {Fenfen Sheng and Zhineng Chen and Bo Xu},
booktitle = {ICDAR},
year = {2019},
url = {http://arxiv.org/abs/1806.00926},
pages = {781-786}
}
</code></pre>
</div>
<div class="row">
<div class="large-6 columns">
<p class="text-left" style="padding:15px 0px;">
</p>
</div>
<div class="large-6 columns">
<p class="text-right" style="padding:15px 0px;">
<a href="17129380736062.html"
title="Next Post: MWeb TestIMage">MWeb TestIMage »</a>
</p>
</div>
</div>
<div class="comments-wrap">
<div class="share-comments">
</div>
</div>
</div><!-- article-wrap -->
</div><!-- large 8 -->
<div class="large-4 medium-4 columns">
<div class="hide-for-small">
<div id="sidebar" class="sidebar">
<div id="site-info" class="site-info">
<h1></h1>
<div class="site-des"></div>
<div class="social">
<a target="_blank" class="rss" href="atom.xml" title="RSS">RSS</a>
</div>
</div>
<div id="site-categories" class="side-item ">
<div class="side-header">
<h2>Categories</h2>
</div>
<div class="side-content">
<p class="cat-list">
</p>
</div>
</div>
<div id="site-categories" class="side-item">
<div class="side-header">
<h2>Recent Posts</h2>
</div>
<div class="side-content">
<ul class="posts-list">
<li class="post">
<a href="17129387431530.html">场景文本识别算法-NRTR</a>
</li>
<li class="post">
<a href="17129380736062.html">MWeb TestIMage</a>
</li>
</ul>
</div>
</div>
</div><!-- sidebar -->
</div><!-- hide for small -->
</div><!-- large 4 -->
</div><!-- row -->
<div class="page-bottom clearfix">
<div class="row">
<p class="copyright">Copyright © 2015
Powered by <a target="_blank" href="http://www.mweb.im">MWeb</a>,
Theme used <a target="_blank" href="http://github.com">GitHub CSS</a>.</p>
</div>
</div>
</section>
</div>
</div>
<style>.mweb-charts{background:#fff;}
body{ box-sizing: border-box;
margin: 0 auto;}
@media print{
pre, code, pre code {
overflow: visible !important;
white-space: pre-wrap !important; /* css-3 */
white-space: -moz-pre-wrap !important; /* Mozilla, since 1999 */
white-space: -pre-wrap !important; /* Opera 4-6 */
white-space: -o-pre-wrap !important; /* Opera 7 */
word-wrap: break-word !important; /* Internet Explorer 5.5+ */
}
html,body{margin:0;padding:4px;}
}
div.code-toolbar {
position: relative;
}
div.code-toolbar > .toolbar {
position: absolute;
z-index: 10;
top: .3em;
right: .2em;
transition: opacity 0.3s ease-in-out;
opacity: 0;
}
div.code-toolbar:hover > .toolbar {
opacity: 1;
}
/* Separate line b/c rules are thrown out if selector is invalid.
IE11 and old Edge versions don't support :focus-within. */
div.code-toolbar:focus-within > .toolbar {
opacity: 1;
}
div.code-toolbar > .toolbar > .toolbar-item {
display: inline-block;
}
div.code-toolbar > .toolbar > .toolbar-item > a {
cursor: pointer;
}
div.code-toolbar > .toolbar > .toolbar-item > button {
background: none;
border: 0;
color: inherit;
font: inherit;
line-height: normal;
overflow: visible;
padding: 0;
-webkit-user-select: none; /* for button */
-moz-user-select: none;
-ms-user-select: none;
}
div.code-toolbar > .toolbar > .toolbar-item > a,
div.code-toolbar > .toolbar > .toolbar-item > button,
div.code-toolbar > .toolbar > .toolbar-item > span {
color: inherit;
font-size: .8em;
padding: 4px .5em;
background: #f5f2f0;
background: rgba(224, 224, 224, 0.4);
box-shadow: 0 2px 0 0 rgba(0,0,0,0.2);
border-radius: .5em;
}
div.code-toolbar > .toolbar > .toolbar-item > a:hover,
div.code-toolbar > .toolbar > .toolbar-item > a:focus,
div.code-toolbar > .toolbar > .toolbar-item > button:hover,
div.code-toolbar > .toolbar > .toolbar-item > button:focus,
div.code-toolbar > .toolbar > .toolbar-item > span:hover,
div.code-toolbar > .toolbar > .toolbar-item > span:focus {
color: inherit;
text-decoration: none;
}
</style><script src="https://cdn.jsdelivr.net/npm/prismjs@1.29.0/prism.min.js"></script><script src="https://cdn.jsdelivr.net/npm/prismjs@1.29.0/plugins/line-numbers/prism-line-numbers.min.js"></script><script>!function(){if("undefined"!=typeof Prism&&"undefined"!=typeof document){var e=[],t={},n=function(){};Prism.plugins.toolbar={};var a=Prism.plugins.toolbar.registerButton=function(n,a){var r;r="function"==typeof a?a:function(e){var t;return"function"==typeof a.onClick?((t=document.createElement("button")).type="button",t.addEventListener("click",(function(){a.onClick.call(this,e)}))):"string"==typeof a.url?(t=document.createElement("a")).href=a.url:t=document.createElement("span"),a.className&&t.classList.add(a.className),t.textContent=a.text,t},n in t?console.warn('There is a button with the key "'+n+'" registered already.'):e.push(t[n]=r)},r=Prism.plugins.toolbar.hook=function(a){var r=a.element.parentNode;var l=a.element.classList;if(l.contains('language-mermaid') || l.contains('language-echarts') || l.contains('language-plantuml')){return;} if(r&&/pre/i.test(r.nodeName)&&!r.parentNode.classList.contains("code-toolbar")){var o=document.createElement("div");o.classList.add("code-toolbar"),r.parentNode.insertBefore(o,r),o.appendChild(r);var i=document.createElement("div");i.classList.add("toolbar");var l=e,d=function(e){for(;e;){var t=e.getAttribute("data-toolbar-order");if(null!=t)return(t=t.trim()).length?t.split(/\s*,\s*/g):[];e=e.parentElement}}(a.element);d&&(l=d.map((function(e){return t[e]||n}))),l.forEach((function(e){var t=e(a);if(t){var n=document.createElement("div");n.classList.add("toolbar-item"),n.appendChild(t),i.appendChild(n)}})),o.appendChild(i)}};a("label",(function(e){var t=e.element.parentNode;if(t&&/pre/i.test(t.nodeName)&&t.hasAttribute("data-label")){var n,a,r=t.getAttribute("data-label");try{a=document.querySelector("template#"+r)}catch(e){}return a?n=a.content:(t.hasAttribute("data-url")?(n=document.createElement("a")).href=t.getAttribute("data-url"):n=document.createElement("span"),n.textContent=r),n}})),Prism.hooks.add("complete",r)}}();</script><link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/prismjs@1.29.0/plugins/toolbar/prism-toolbar.min.css"><script src="https://cdn.jsdelivr.net/npm/prismjs@1.29.0/plugins/copy-to-clipboard/prism-copy-to-clipboard.min.js"></script><script src="https://cdn.jsdelivr.net/npm/prismjs@1.29.0/plugins/autoloader/prism-autoloader.min.js"></script><style>div.code-toolbar > .toolbar > .toolbar-item > a, div.code-toolbar > .toolbar > .toolbar-item > button, div.code-toolbar > .toolbar > .toolbar-item > span {padding: 4px .5em; background: #f5f2f0; background: rgba(224, 224, 224, 0.4);}</style>
<style type="text/css">
figure{margin: 0;padding: 0;}
figcaption{text-align:center;}
/* PrismJS 1.14.0
http://prismjs.com/download.html#themes=prism&languages=markup+css+clike+javascript */
/**
* prism.js default theme for JavaScript, CSS and HTML
* Based on dabblet (http://dabblet.com)
* @author Lea Verou
*/
code[class*="language-"],
pre[class*="language-"] {
color: black;
background: none;
text-shadow: 0 1px white;
font-family: Consolas, Monaco, 'Andale Mono', 'Ubuntu Mono', monospace;
text-align: left;
white-space: pre;
word-spacing: normal;
word-break: normal;
word-wrap: normal;
line-height: 1.5;
-moz-tab-size: 4;
-o-tab-size: 4;
tab-size: 4;
-webkit-hyphens: none;
-moz-hyphens: none;
-ms-hyphens: none;
hyphens: none;
}
pre[class*="language-"]::-moz-selection, pre[class*="language-"] ::-moz-selection,
code[class*="language-"]::-moz-selection, code[class*="language-"] ::-moz-selection {
text-shadow: none;
background:#b3d4fc;
}
pre[class*="language-"]::selection, pre[class*="language-"] ::selection,
code[class*="language-"]::selection, code[class*="language-"] ::selection {
text-shadow: none;
background: #b3d4fc;
}
@media print {
code[class*="language-"],
pre[class*="language-"] {
text-shadow: none;
}
}
/* Code blocks */
pre[class*="language-"] {
padding: 1em;
margin: .5em 0;
overflow: auto;
}
:not(pre) > code[class*="language-"],
pre[class*="language-"] {
background: #F7F7F7;
}
/* Inline code */
:not(pre) > code[class*="language-"] {
padding: .1em;
border-radius: .3em;
white-space: normal;
}
.token.comment,
.token.prolog,
.token.doctype,
.token.cdata {
color: slategray;
}
.token.punctuation {
color: #999;
}
.namespace {
opacity: .7;
}
.token.property,
.token.tag,
.token.boolean,
.token.number,
.token.constant,
.token.symbol,
.token.deleted {
color: #905;
}
.token.selector,
.token.attr-name,
.token.string,
.token.char,
.token.builtin,
.token.inserted {
color: #690;
}
.token.operator,
.token.entity,
.token.url,
.language-css .token.string,
.style .token.string {
color: #9a6e3a;
background: hsla(0, 0%, 100%, .5);
}
.token.atrule,
.token.attr-value,
.token.keyword {
color: #07a;
}
.token.function,
.token.class-name {
color: #DD4A68;
}
.token.regex,
.token.important,
.token.variable {
color: #e90;
}
.token.important,
.token.bold {
font-weight: bold;
}
.token.italic {
font-style: italic;
}
.token.entity {
cursor: help;
}
pre[class*="language-"].line-numbers {
position: relative;
padding-left: 3.8em;
counter-reset: linenumber;
}
pre[class*="language-"].line-numbers > code {
position: relative;
white-space: inherit;
}
.line-numbers .line-numbers-rows {
position: absolute;
pointer-events: none;
top: 0;
font-size: 100%;
left: -3.8em;
width: 3em; /* works for line-numbers below 1000 lines */
letter-spacing: -1px;
border-right: 1px solid #999;
-webkit-user-select: none;
-moz-user-select: none;
-ms-user-select: none;
user-select: none;
}
.line-numbers-rows > span {
pointer-events: none;
display: block;
counter-increment: linenumber;
}
.line-numbers-rows > span:before {
content: counter(linenumber);
color: #999;
display: block;
padding-right: 0.8em;
text-align: right;
}
</style>
<script src="asset/js/foundation.min.js"></script>
<script>
$(document).foundation();
function fixSidebarHeight(){
var w1 = $('.markdown-body').height();
var w2 = $('#sidebar').height();
if (w1 > w2) { $('#sidebar').height(w1); };
}
$(function(){
fixSidebarHeight();
})
$(window).load(function(){
fixSidebarHeight();
});
</script>
</body>
</html>