recognition_en.md 18.5 KB
Newer Older
tink2123's avatar
tink2123 committed
1
# TEXT RECOGNITION
Khanh Tran's avatar
Khanh Tran committed
2

WenmuZhou's avatar
WenmuZhou committed
3
4
5
6
7
- [1 DATA PREPARATION](#DATA_PREPARATION)
    - [1.1 Costom Dataset](#Costom_Dataset)
    - [1.2 Dataset Download](#Dataset_download)
    - [1.3 Dictionary](#Dictionary)  
    - [1.4 Add Space Category](#Add_space_category)
WenmuZhou's avatar
WenmuZhou committed
8

WenmuZhou's avatar
WenmuZhou committed
9
10
- [2 TRAINING](#TRAINING)
    - [2.1 Data Augmentation](#Data_Augmentation)
tink2123's avatar
tink2123 committed
11
12
    - [2.2 General Training](#Training)
    - [2.3 Multi-language Training](#Multi_language)
WenmuZhou's avatar
WenmuZhou committed
13

WenmuZhou's avatar
WenmuZhou committed
14
- [3 EVALUATION](#EVALUATION)
WenmuZhou's avatar
WenmuZhou committed
15

WenmuZhou's avatar
WenmuZhou committed
16
17
- [4 PREDICTION](#PREDICTION)
    - [4.1 Training engine prediction](#Training_engine_prediction)
18
- [5 CONVERT TO INFERENCE MODEL](#Inference)
WenmuZhou's avatar
WenmuZhou committed
19
20

<a name="DATA_PREPARATION"></a>
tink2123's avatar
tink2123 committed
21
## 1 DATA PREPARATION
Khanh Tran's avatar
Khanh Tran committed
22
23


WenmuZhou's avatar
WenmuZhou committed
24
PaddleOCR supports two data formats:
tink2123's avatar
tink2123 committed
25
26
- `LMDB` is used to train data sets stored in lmdb format(LMDBDataSet);
- `general data` is used to train data sets stored in text files(SimpleDataSet):
Khanh Tran's avatar
Khanh Tran committed
27
28
29
30
31
32

Please organize the dataset as follows:

The default storage path for training data is `PaddleOCR/train_data`, if you already have a dataset on your disk, just create a soft link to the dataset directory:

```
WenmuZhou's avatar
WenmuZhou committed
33
# linux and mac os
34
ln -sf <path/to/dataset> <path/to/paddle_ocr>/train_data/dataset
WenmuZhou's avatar
WenmuZhou committed
35
36
# windows
mklink /d <path/to/paddle_ocr>/train_data/dataset <path/to/dataset>
Khanh Tran's avatar
Khanh Tran committed
37
38
```

WenmuZhou's avatar
WenmuZhou committed
39
<a name="Costom_Dataset"></a>
tink2123's avatar
tink2123 committed
40
### 1.1 Costom dataset
Khanh Tran's avatar
Khanh Tran committed
41
42
43
44
45

If you want to use your own data for training, please refer to the following to organize your data.

- Training set

WenmuZhou's avatar
WenmuZhou committed
46
It is recommended to put the training images in the same folder, and use a txt file (rec_gt_train.txt) to store the image path and label. The contents of the txt file are as follows:
Khanh Tran's avatar
Khanh Tran committed
47
48
49
50
51
52

* Note: by default, the image path and image label are split with \t, if you use other methods to split, it will cause training error

```
" Image file name           Image annotation "

WenmuZhou's avatar
WenmuZhou committed
53
54
train_data/rec/train/word_001.jpg   简单可依赖
train_data/rec/train/word_002.jpg   用科技让复杂的世界更简单
WenmuZhou's avatar
WenmuZhou committed
55
...
Khanh Tran's avatar
Khanh Tran committed
56
57
58
59
60
61
```

The final training set should have the following file structure:

```
|-train_data
WenmuZhou's avatar
WenmuZhou committed
62
  |-rec
WenmuZhou's avatar
WenmuZhou committed
63
64
65
66
67
68
    |- rec_gt_train.txt
    |- train
        |- word_001.png
        |- word_002.jpg
        |- word_003.jpg
        | ...
Khanh Tran's avatar
Khanh Tran committed
69
70
71
72
73
74
75
76
```

- Test set

Similar to the training set, the test set also needs to be provided a folder containing all images (test) and a rec_gt_test.txt. The structure of the test set is as follows:

```
|-train_data
WenmuZhou's avatar
WenmuZhou committed
77
  |-rec
Khanh Tran's avatar
Khanh Tran committed
78
79
80
81
82
83
84
85
    |-ic15_data
        |- rec_gt_test.txt
        |- test
            |- word_001.jpg
            |- word_002.jpg
            |- word_003.jpg
            | ...
```
WenmuZhou's avatar
WenmuZhou committed
86
87

<a name="Dataset_download"></a>
tink2123's avatar
tink2123 committed
88
### 1.2 Dataset download
WenmuZhou's avatar
WenmuZhou committed
89

tink2123's avatar
tink2123 committed
90
- ICDAR2015
WenmuZhou's avatar
WenmuZhou committed
91

tink2123's avatar
tink2123 committed
92
93
If you do not have a dataset locally, you can download it on the official website [icdar2015](http://rrc.cvc.uab.es/?ch=4&com=downloads).
Also refer to [DTRB](https://github.com/clovaai/deep-text-recognition-benchmark#download-lmdb-dataset-for-traininig-and-evaluation-from-here) ,download the lmdb format dataset required for benchmark
WenmuZhou's avatar
WenmuZhou committed
94
95
96
97
98
99
100
101
102
103

PaddleOCR provides label files for training the icdar2015 dataset, which can be downloaded in the following ways:

```
# Training set label
wget -P ./train_data/ic15_data  https://paddleocr.bj.bcebos.com/dataset/rec_gt_train.txt
# Test Set Label
wget -P ./train_data/ic15_data  https://paddleocr.bj.bcebos.com/dataset/rec_gt_test.txt
```

tink2123's avatar
tink2123 committed
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
PaddleOCR also provides a data format conversion script, which can convert ICDAR official website label to a data format
supported by PaddleOCR. The data conversion tool is in `ppocr/utils/gen_label.py`, here is the training set as an example:

```
# convert the official gt to rec_gt_label.txt
python gen_label.py --mode="rec" --input_path="{path/of/origin/label}" --output_label="rec_gt_label.txt"
```

The data format is as follows, (a) is the original picture, (b) is the Ground Truth text file corresponding to each picture:

![](../datasets/icdar_rec.png)


- Multilingual dataset

The multi-language model training method is the same as the Chinese model. The training data set is 100w synthetic data. A small amount of fonts and test data can be downloaded using the following two methods.
* [Baidu Netdisk](https://pan.baidu.com/s/1bS_u207Rm7YbY33wOECKDA) ,Extraction code:frgi.
* [Google drive](https://drive.google.com/file/d/18cSWX7wXSy4G0tbKJ0d9PuIaiwRLHpjA/view)


WenmuZhou's avatar
WenmuZhou committed
124
<a name="Dictionary"></a>
tink2123's avatar
tink2123 committed
125
### 1.3 Dictionary
Khanh Tran's avatar
Khanh Tran committed
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141

Finally, a dictionary ({word_dict_name}.txt) needs to be provided so that when the model is trained, all the characters that appear can be mapped to the dictionary index.

Therefore, the dictionary needs to contain all the characters that you want to be recognized correctly. {word_dict_name}.txt needs to be written in the following format and saved in the `utf-8` encoding format:

```
l
d
a
d
r
n
```

In `word_dict.txt`, there is a single word in each line, which maps characters and numeric indexes together, e.g "and" will be mapped to [2 5 1]

WenmuZhou's avatar
WenmuZhou committed
142
143
PaddleOCR has built-in dictionaries, which can be used on demand.

Khanh Tran's avatar
Khanh Tran committed
144
145
`ppocr/utils/ppocr_keys_v1.txt` is a Chinese dictionary with 6623 characters.

WenmuZhou's avatar
WenmuZhou committed
146
147
148
149
`ppocr/utils/ic15_dict.txt` is an English dictionary with 63 characters

`ppocr/utils/dict/french_dict.txt` is a French dictionary with 118 characters

150
`ppocr/utils/dict/japan_dict.txt` is a Japanese dictionary with 4399 characters
WenmuZhou's avatar
WenmuZhou committed
151

tink2123's avatar
tink2123 committed
152
`ppocr/utils/dict/korean_dict.txt` is a Korean dictionary with 3636 characters
WenmuZhou's avatar
WenmuZhou committed
153

tink2123's avatar
tink2123 committed
154
155
`ppocr/utils/dict/german_dict.txt` is a German dictionary with 131 characters

tink2123's avatar
tink2123 committed
156
`ppocr/utils/en_dict.txt` is a English dictionary with 96 characters
WenmuZhou's avatar
WenmuZhou committed
157

158

WenmuZhou's avatar
WenmuZhou committed
159
The current multi-language model is still in the demo stage and will continue to optimize the model and add languages. **You are very welcome to provide us with dictionaries and fonts in other languages**,
littletomatodonkey's avatar
fix doc  
littletomatodonkey committed
160
If you like, you can submit the dictionary file to [dict](../../ppocr/utils/dict) and we will thank you in the Repo.
Khanh Tran's avatar
Khanh Tran committed
161
162
163
164


To customize the dict file, please modify the `character_dict_path` field in `configs/rec/rec_icdar15_train.yml` and set `character_type` to `ch`.

tink2123's avatar
tink2123 committed
165
166
167
168
- Custom dictionary

If you need to customize dic file, please add character_dict_path field in configs/rec/rec_icdar15_train.yml to point to your dictionary path. And set character_type to ch.

WenmuZhou's avatar
WenmuZhou committed
169
<a name="Add_space_category"></a>
tink2123's avatar
tink2123 committed
170
### 1.4 Add space category
tink2123's avatar
tink2123 committed
171

xmy0916's avatar
xmy0916 committed
172
If you want to support the recognition of the `space` category, please set the `use_space_char` field in the yml file to `True`.
tink2123's avatar
tink2123 committed
173
174
175

**Note: use_space_char only takes effect when character_type=ch**

WenmuZhou's avatar
WenmuZhou committed
176
<a name="TRAINING"></a>
tink2123's avatar
tink2123 committed
177
## 2 TRAINING
Khanh Tran's avatar
Khanh Tran committed
178

tink2123's avatar
tink2123 committed
179
<a name="Data_Augmentation"></a>
tink2123's avatar
tink2123 committed
180
### 2.1 Data Augmentation
tink2123's avatar
tink2123 committed
181
182
183
184
185
186
187
188

PaddleOCR provides a variety of data augmentation methods. All the augmentation methods are enabled by default.

The default perturbation methods are: cvtColor, blur, jitter, Gasuss noise, random crop, perspective, color reverse, TIA augmentation.

Each disturbance method is selected with a 40% probability during the training process. For specific code implementation, please refer to: [rec_img_aug.py](../../ppocr/data/imaug/rec_img_aug.py)

<a name="Training"></a>
tink2123's avatar
tink2123 committed
189
### 2.2 General Training
tink2123's avatar
tink2123 committed
190

Khanh Tran's avatar
Khanh Tran committed
191
192
193
194
195
196
197
PaddleOCR provides training scripts, evaluation scripts, and prediction scripts. In this section, the CRNN recognition model will be used as an example:

First download the pretrain model, you can download the trained model to finetune on the icdar2015 data:

```
cd PaddleOCR/
# Download the pre-trained model of MobileNetV3
tink2123's avatar
tink2123 committed
198
wget -P ./pretrain_models/ https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_mv3_none_bilstm_ctc_v2.0_train.tar
Khanh Tran's avatar
Khanh Tran committed
199
200
# Decompress model parameters
cd pretrain_models
tink2123's avatar
tink2123 committed
201
tar -xf rec_mv3_none_bilstm_ctc_v2.0_train.tar && rm -rf rec_mv3_none_bilstm_ctc_v2.0_train.tar
Khanh Tran's avatar
Khanh Tran committed
202
203
204
205
206
```

Start training:

```
tink2123's avatar
tink2123 committed
207
# GPU training Support single card and multi-card training
tink2123's avatar
tink2123 committed
208
# Training icdar15 English data and The training log will be automatically saved as train.log under "{save_model_dir}"
tink2123's avatar
tink2123 committed
209
210
211
212

#specify the single card training(Long training time, not recommended)
python3 tools/train.py -c configs/rec/rec_icdar15_train.yml
#specify the card number through --gpus
xmy0916's avatar
xmy0916 committed
213
python3 -m paddle.distributed.launch --gpus '0,1,2,3'  tools/train.py -c configs/rec/rec_icdar15_train.yml
Khanh Tran's avatar
Khanh Tran committed
214
```
tink2123's avatar
tink2123 committed
215
216


Khanh Tran's avatar
Khanh Tran committed
217
218
219
220
221
222
223
224
225
PaddleOCR supports alternating training and evaluation. You can modify `eval_batch_step` in `configs/rec/rec_icdar15_train.yml` to set the evaluation frequency. By default, it is evaluated every 500 iter and the best acc model is saved under `output/rec_CRNN/best_accuracy` during the evaluation process.

If the evaluation set is large, the test will be time-consuming. It is recommended to reduce the number of evaluations, or evaluate after training.

* Tip: You can use the `-c` parameter to select multiple model configurations under the `configs/rec/` path for training. The recognition algorithms supported by PaddleOCR are:


| Configuration file |  Algorithm |   backbone |   trans   |   seq      |     pred     |
| :--------: |  :-------:   | :-------:  |   :-------:   |   :-----:   |  :-----:   |
xmy0916's avatar
xmy0916 committed
226
227
| [rec_chinese_lite_train_v2.0.yml](../../configs/rec/ch_ppocr_v2.0/rec_chinese_lite_train_v2.0.yml) |  CRNN |   Mobilenet_v3 small 0.5 |  None   |  BiLSTM |  ctc  |
| [rec_chinese_common_train_v2.0.yml](../../configs/rec/ch_ppocr_v2.0/rec_chinese_common_train_v2.0.yml) |  CRNN | ResNet34_vd |  None   |  BiLSTM |  ctc  |
Khanh Tran's avatar
Khanh Tran committed
228
| rec_chinese_lite_train.yml |  CRNN |   Mobilenet_v3 small 0.5 |  None   |  BiLSTM |  ctc  |
WenmuZhou's avatar
WenmuZhou committed
229
| rec_chinese_common_train.yml |  CRNN |   ResNet34_vd |  None   |  BiLSTM |  ctc  |
Khanh Tran's avatar
Khanh Tran committed
230
231
232
233
234
| rec_icdar15_train.yml |  CRNN |   Mobilenet_v3 large 0.5 |  None   |  BiLSTM |  ctc  |
| rec_mv3_none_bilstm_ctc.yml |  CRNN |   Mobilenet_v3 large 0.5 |  None   |  BiLSTM |  ctc  |
| rec_mv3_none_none_ctc.yml |  Rosetta |   Mobilenet_v3 large 0.5 |  None   |  None |  ctc  |
| rec_r34_vd_none_bilstm_ctc.yml |  CRNN |   Resnet34_vd |  None   |  BiLSTM |  ctc  |
| rec_r34_vd_none_none_ctc.yml |  Rosetta |   Resnet34_vd |  None   |  None |  ctc  |
LDOUBLEV's avatar
LDOUBLEV committed
235
236
| rec_mv3_tps_bilstm_att.yml |  CRNN |   Mobilenet_v3 |  TPS   |  BiLSTM |  att  |
| rec_r34_vd_tps_bilstm_att.yml |  CRNN |   Resnet34_vd |  TPS   |  BiLSTM |  att  |
tink2123's avatar
tink2123 committed
237
| rec_r50fpn_vd_none_srn.yml    | SRN | Resnet50_fpn_vd    | None    | rnn | srn |
Topdu's avatar
Topdu committed
238
| rec_mtb_nrtr.yml    | NRTR | nrtr_mtb    | None    | transformer encoder | transformer decoder |
Khanh Tran's avatar
Khanh Tran committed
239

WenmuZhou's avatar
WenmuZhou committed
240
For training Chinese data, it is recommended to use
xmy0916's avatar
xmy0916 committed
241
[rec_chinese_lite_train_v2.0.yml](../../configs/rec/ch_ppocr_v2.0/rec_chinese_lite_train_v2.0.yml). If you want to try the result of other algorithms on the Chinese data set, please refer to the following instructions to modify the configuration file:
Khanh Tran's avatar
Khanh Tran committed
242
co
xmy0916's avatar
xmy0916 committed
243
Take `rec_chinese_lite_train_v2.0.yml` as an example:
Khanh Tran's avatar
Khanh Tran committed
244
245
246
```
Global:
  ...
xmy0916's avatar
xmy0916 committed
247
248
  # Add a custom dictionary, such as modify the dictionary, please point the path to the new dictionary
  character_dict_path: ppocr/utils/ppocr_keys_v1.txt
Khanh Tran's avatar
Khanh Tran committed
249
250
251
  # Modify character type
  character_type: ch
  ...
252
  # Whether to recognize spaces
xmy0916's avatar
xmy0916 committed
253
  use_space_char: True
Khanh Tran's avatar
Khanh Tran committed
254

255
256
257
258

Optimizer:
  ...
  # Add learning rate decay strategy
xmy0916's avatar
xmy0916 committed
259
260
261
262
263
264
265
266
267
  lr:
    name: Cosine
    learning_rate: 0.001
  ...

...

Train:
  dataset:
MissPenguin's avatar
MissPenguin committed
268
    # Type of dataset,we support LMDBDataSet and SimpleDataSet
xmy0916's avatar
xmy0916 committed
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
    name: SimpleDataSet
    # Path of dataset
    data_dir: ./train_data/
    # Path of train list
    label_file_list: ["./train_data/train_list.txt"]
    transforms:
      ...
      - RecResizeImg:
          # Modify image_shape to fit long text
          image_shape: [3, 32, 320]
      ...
  loader:
    ...
    # Train batch_size for Single card
    batch_size_per_card: 256
    ...

Eval:
  dataset:
MissPenguin's avatar
MissPenguin committed
288
    # Type of dataset,we support LMDBDataSet and SimpleDataSet
xmy0916's avatar
xmy0916 committed
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
    name: SimpleDataSet
    # Path of dataset
    data_dir: ./train_data
    # Path of eval list
    label_file_list: ["./train_data/val_list.txt"]
    transforms:
      ...
      - RecResizeImg:
          # Modify image_shape to fit long text
          image_shape: [3, 32, 320]
      ...
  loader:
    # Eval batch_size for Single card
    batch_size_per_card: 256
    ...
Khanh Tran's avatar
Khanh Tran committed
304
305
306
```
**Note that the configuration file for prediction/evaluation must be consistent with the training.**

WenmuZhou's avatar
WenmuZhou committed
307
<a name="Multi_language"></a>
tink2123's avatar
tink2123 committed
308
### 2.3 Multi-language Training
tink2123's avatar
tink2123 committed
309
310
311

Currently, the multi-language algorithms supported by PaddleOCR are:

tink2123's avatar
tink2123 committed
312
| Configuration file |  Algorithm name |   backbone |   trans   |   seq      |     pred     |  language | character_type |
tink2123's avatar
tink2123 committed
313
314
| :--------: |  :-------:   | :-------:  |   :-------:   |   :-----:   |  :-----:   | :-----:  | :-----:  |
| rec_chinese_cht_lite_train.yml |  CRNN |   Mobilenet_v3 small 0.5 |  None   |  BiLSTM |  ctc  | chinese traditional  | chinese_cht|
tink2123's avatar
tink2123 committed
315
| rec_en_lite_train.yml |  CRNN |   Mobilenet_v3 small 0.5 |  None   |  BiLSTM |  ctc  | English(Case sensitive)   | EN |
tink2123's avatar
tink2123 committed
316
317
318
319
| rec_french_lite_train.yml |  CRNN |   Mobilenet_v3 small 0.5 |  None   |  BiLSTM |  ctc  | French |  french |
| rec_ger_lite_train.yml |  CRNN |   Mobilenet_v3 small 0.5 |  None   |  BiLSTM |  ctc  | German   | german |
| rec_japan_lite_train.yml |  CRNN |   Mobilenet_v3 small 0.5 |  None   |  BiLSTM |  ctc  | Japanese | japan |
| rec_korean_lite_train.yml |  CRNN |   Mobilenet_v3 small 0.5 |  None   |  BiLSTM |  ctc  | Korean  | korean |
tink2123's avatar
tink2123 committed
320
321
322
323
| rec_latin_lite_train.yml |  CRNN |   Mobilenet_v3 small 0.5 |  None   |  BiLSTM |  ctc  | Latin  | latin |
| rec_arabic_lite_train.yml |  CRNN |   Mobilenet_v3 small 0.5 |  None   |  BiLSTM |  ctc  | arabic |  ar |
| rec_cyrillic_lite_train.yml |  CRNN |   Mobilenet_v3 small 0.5 |  None   |  BiLSTM |  ctc  | cyrillic   | cyrillic |
| rec_devanagari_lite_train.yml |  CRNN |   Mobilenet_v3 small 0.5 |  None   |  BiLSTM |  ctc  | devanagari  | devanagari |
tink2123's avatar
tink2123 committed
324

tink2123's avatar
tink2123 committed
325
For more supported languages, please refer to : [Multi-language model](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.1/doc/doc_en/multi_languages_en.md#4-support-languages-and-abbreviations)
WenmuZhou's avatar
WenmuZhou committed
326
327
328
329
330
331
332
333
334


If you want to finetune on the basis of the existing model effect, please refer to the following instructions to modify the configuration file:

Take `rec_french_lite_train` as an example:

```
Global:
  ...
xmy0916's avatar
xmy0916 committed
335
  # Add a custom dictionary, such as modify the dictionary, please point the path to the new dictionary
WenmuZhou's avatar
WenmuZhou committed
336
337
  character_dict_path: ./ppocr/utils/dict/french_dict.txt
  ...
xmy0916's avatar
xmy0916 committed
338
  # Whether to recognize spaces
xmy0916's avatar
xmy0916 committed
339
  use_space_char: True
xmy0916's avatar
xmy0916 committed
340

WenmuZhou's avatar
WenmuZhou committed
341
...
xmy0916's avatar
xmy0916 committed
342
343
344

Train:
  dataset:
MissPenguin's avatar
MissPenguin committed
345
    # Type of dataset,we support LMDBDataSet and SimpleDataSet
xmy0916's avatar
xmy0916 committed
346
347
348
349
350
351
352
353
354
    name: SimpleDataSet
    # Path of dataset
    data_dir: ./train_data/
    # Path of train list
    label_file_list: ["./train_data/french_train.txt"]
    ...

Eval:
  dataset:
MissPenguin's avatar
MissPenguin committed
355
    # Type of dataset,we support LMDBDataSet and SimpleDataSet
xmy0916's avatar
xmy0916 committed
356
357
358
359
360
361
    name: SimpleDataSet
    # Path of dataset
    data_dir: ./train_data
    # Path of eval list
    label_file_list: ["./train_data/french_val.txt"]
    ...
WenmuZhou's avatar
WenmuZhou committed
362
```
Khanh Tran's avatar
Khanh Tran committed
363

WenmuZhou's avatar
WenmuZhou committed
364
<a name="EVALUATION"></a>
365

tink2123's avatar
tink2123 committed
366
## 3 EVALUATION
Khanh Tran's avatar
Khanh Tran committed
367

WenmuZhou's avatar
WenmuZhou committed
368
The evaluation dataset can be set by modifying the `Eval.dataset.label_file_list` field in the `configs/rec/rec_icdar15_train.yml` file.
Khanh Tran's avatar
Khanh Tran committed
369
370
371

```
# GPU evaluation, Global.checkpoints is the weight to be tested
WenmuZhou's avatar
WenmuZhou committed
372
python3 -m paddle.distributed.launch --gpus '0' tools/eval.py -c configs/rec/rec_icdar15_train.yml -o Global.checkpoints={path/to/weights}/best_accuracy
Khanh Tran's avatar
Khanh Tran committed
373
374
```

WenmuZhou's avatar
WenmuZhou committed
375
<a name="PREDICTION"></a>
tink2123's avatar
tink2123 committed
376
## 4 PREDICTION
Khanh Tran's avatar
Khanh Tran committed
377
378
379
380


Using the model trained by paddleocr, you can quickly get prediction through the following script.

tink2123's avatar
tink2123 committed
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
The default prediction picture is stored in `infer_img`, and the trained weight is specified via `-o Global.checkpoints`:


According to the `save_model_dir` and `save_epoch_step` fields set in the configuration file, the following parameters will be saved:

```
output/rec/
├── best_accuracy.pdopt  
├── best_accuracy.pdparams  
├── best_accuracy.states  
├── config.yml  
├── iter_epoch_3.pdopt  
├── iter_epoch_3.pdparams  
├── iter_epoch_3.states  
├── latest.pdopt  
├── latest.pdparams  
├── latest.states  
└── train.log
```

Among them, best_accuracy.* is the best model on the evaluation set; iter_epoch_x.* is the model saved at intervals of `save_epoch_step`; latest.* is the model of the last epoch.
Khanh Tran's avatar
Khanh Tran committed
402
403
404

```
# Predict English results
WenmuZhou's avatar
WenmuZhou committed
405
python3 tools/infer_rec.py -c configs/rec/ch_ppocr_v2.0/rec_chinese_lite_train_v2.0.yml -o Global.pretrained_model={path/to/weights}/best_accuracy Global.load_static_weights=false Global.infer_img=doc/imgs_words/en/word_1.jpg
Khanh Tran's avatar
Khanh Tran committed
406
407
```

tink2123's avatar
tink2123 committed
408

Khanh Tran's avatar
Khanh Tran committed
409
410
Input image:

411
![](../imgs_words/en/word_1.png)
Khanh Tran's avatar
Khanh Tran committed
412
413
414
415
416

Get the prediction result of the input image:

```
infer_img: doc/imgs_words/en/word_1.png
tink2123's avatar
tink2123 committed
417
        result: ('joint', 0.9998967)
Khanh Tran's avatar
Khanh Tran committed
418
419
```

xmy0916's avatar
xmy0916 committed
420
The configuration file used for prediction must be consistent with the training. For example, you completed the training of the Chinese model with `python3 tools/train.py -c configs/rec/ch_ppocr_v2.0/rec_chinese_lite_train_v2.0.yml`, you can use the following command to predict the Chinese model:
Khanh Tran's avatar
Khanh Tran committed
421
422
423

```
# Predict Chinese results
WenmuZhou's avatar
WenmuZhou committed
424
python3 tools/infer_rec.py -c configs/rec/ch_ppocr_v2.0/rec_chinese_lite_train_v2.0.yml -o Global.pretrained_model={path/to/weights}/best_accuracy Global.load_static_weights=false Global.infer_img=doc/imgs_words/ch/word_1.jpg
Khanh Tran's avatar
Khanh Tran committed
425
426
427
428
```

Input image:

429
![](../imgs_words/ch/word_1.jpg)
Khanh Tran's avatar
Khanh Tran committed
430
431
432
433
434

Get the prediction result of the input image:

```
infer_img: doc/imgs_words/ch/word_1.jpg
tink2123's avatar
tink2123 committed
435
        result: ('韩国小馆', 0.997218)
Khanh Tran's avatar
Khanh Tran committed
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

<a name="Inference"></a>

## 5 CONVERT TO INFERENCE MODEL

The recognition model is converted to the inference model in the same way as the detection, as follows:

```
# -c Set the training algorithm yml configuration file
# -o Set optional parameters
# Global.pretrained_model parameter Set the training model address to be converted without adding the file suffix .pdmodel, .pdopt or .pdparams.
# Global.save_inference_dir Set the address where the converted model will be saved.

python3 tools/export_model.py -c configs/rec/ch_ppocr_v2.0/rec_chinese_lite_train_v2.0.yml -o Global.pretrained_model=./ch_lite/ch_ppocr_mobile_v2.0_rec_train/best_accuracy  Global.save_inference_dir=./inference/rec_crnn/
```

If you have a model trained on your own dataset with a different dictionary file, please make sure that you modify the `character_dict_path` in the configuration file to your dictionary file path.

After the conversion is successful, there are three files in the model save directory:

```
inference/det_db/
    ├── inference.pdiparams         # The parameter file of recognition inference model
    ├── inference.pdiparams.info    # The parameter information of recognition inference model, which can be ignored
    └── inference.pdmodel           # The program file of recognition model
```

- Text recognition model Inference using custom characters dictionary

  If the text dictionary is modified during training, when using the inference model to predict, you need to specify the dictionary path used by `--rec_char_dict_path`, and set `rec_char_type=ch`

  ```
  python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words_en/word_336.png" --rec_model_dir="./your inference model" --rec_image_shape="3, 32, 100" --rec_char_type="ch" --rec_char_dict_path="your text dict path"
  ```