recognition_en.md 15.5 KB
Newer Older
1
# Text Recognition
Khanh Tran's avatar
Khanh Tran committed
2

WenmuZhou's avatar
WenmuZhou committed
3
4
5
6
7
8
9
10
11
12
13
14
15
- [Text Recognition](#text-recognition)
  - [1. Data Preparation](#1-data-preparation)
    - [1.1 DataSet Preparation](#11-dataset-preparation)
    - [1.2 Dictionary](#12-dictionary)
    - [1.4 Add Space Category](#14-add-space-category)
  - [2.Training](#2training)
    - [2.1 Data Augmentation](#21-data-augmentation)
    - [2.2 General Training](#22-general-training)
    - [2.3 Multi-language Training](#23-multi-language-training)
    - [2.4 Training with Knowledge Distillation](#24-training-with-knowledge-distillation)
  - [3. Evalution](#3-evalution)
  - [4. Prediction](#4-prediction)
  - [5. Convert to Inference Model](#5-convert-to-inference-model)
WenmuZhou's avatar
WenmuZhou committed
16
17

<a name="DATA_PREPARATION"></a>
18
## 1. Data Preparation
Khanh Tran's avatar
Khanh Tran committed
19

WenmuZhou's avatar
WenmuZhou committed
20
### 1.1 DataSet Preparation
Khanh Tran's avatar
Khanh Tran committed
21

WenmuZhou's avatar
WenmuZhou committed
22
To prepare datasets, refer to [ocr_datasets](./dataset/ocr_datasets.md) .
WenmuZhou's avatar
WenmuZhou committed
23

24
25
If you want to reproduce the paper SAR, you need to download extra dataset [SynthAdd](https://pan.baidu.com/share/init?surl=uV0LtoNmcxbO-0YA7Ch4dg), extraction code: 627x. Besides, icdar2013, icdar2015, cocotext, IIIT5k datasets are also used to train. For specific details, please refer to the paper SAR.

WenmuZhou's avatar
WenmuZhou committed
26
<a name="Dictionary"></a>
WenmuZhou's avatar
WenmuZhou committed
27
### 1.2 Dictionary
Khanh Tran's avatar
Khanh Tran committed
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43

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
44
45
PaddleOCR has built-in dictionaries, which can be used on demand.

Khanh Tran's avatar
Khanh Tran committed
46
47
`ppocr/utils/ppocr_keys_v1.txt` is a Chinese dictionary with 6623 characters.

WenmuZhou's avatar
WenmuZhou committed
48
49
50
51
`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

52
`ppocr/utils/dict/japan_dict.txt` is a Japanese dictionary with 4399 characters
WenmuZhou's avatar
WenmuZhou committed
53

tink2123's avatar
tink2123 committed
54
`ppocr/utils/dict/korean_dict.txt` is a Korean dictionary with 3636 characters
WenmuZhou's avatar
WenmuZhou committed
55

tink2123's avatar
tink2123 committed
56
57
`ppocr/utils/dict/german_dict.txt` is a German dictionary with 131 characters

tink2123's avatar
tink2123 committed
58
`ppocr/utils/en_dict.txt` is a English dictionary with 96 characters
WenmuZhou's avatar
WenmuZhou committed
59

60

WenmuZhou's avatar
WenmuZhou committed
61
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
62
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
63
64


tink2123's avatar
tink2123 committed
65
To customize the dict file, please modify the `character_dict_path` field in `configs/rec/rec_icdar15_train.yml` .
Khanh Tran's avatar
Khanh Tran committed
66

tink2123's avatar
tink2123 committed
67
68
69
70
- 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
71
<a name="Add_space_category"></a>
72
### 1.4 Add Space Category
tink2123's avatar
tink2123 committed
73

xmy0916's avatar
xmy0916 committed
74
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
75

WenmuZhou's avatar
WenmuZhou committed
76
<a name="TRAINING"></a>
77
## 2.Training
Khanh Tran's avatar
Khanh Tran committed
78

tink2123's avatar
tink2123 committed
79
<a name="Data_Augmentation"></a>
tink2123's avatar
tink2123 committed
80
### 2.1 Data Augmentation
tink2123's avatar
tink2123 committed
81
82
83
84
85
86
87
88

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
89
### 2.2 General Training
tink2123's avatar
tink2123 committed
90

Khanh Tran's avatar
Khanh Tran committed
91
92
93
94
95
96
97
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
98
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
99
100
# Decompress model parameters
cd pretrain_models
tink2123's avatar
tink2123 committed
101
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
102
103
104
105
106
```

Start training:

```
tink2123's avatar
tink2123 committed
107
# GPU training Support single card and multi-card training
tink2123's avatar
tink2123 committed
108
# Training icdar15 English data and The training log will be automatically saved as train.log under "{save_model_dir}"
tink2123's avatar
tink2123 committed
109
110
111
112

#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
113
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
114
```
tink2123's avatar
tink2123 committed
115
116


Khanh Tran's avatar
Khanh Tran committed
117
118
119
120
121
122
123
124
125
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
126
127
| [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
128
| rec_chinese_lite_train.yml |  CRNN |   Mobilenet_v3 small 0.5 |  None   |  BiLSTM |  ctc  |
WenmuZhou's avatar
WenmuZhou committed
129
| rec_chinese_common_train.yml |  CRNN |   ResNet34_vd |  None   |  BiLSTM |  ctc  |
Khanh Tran's avatar
Khanh Tran committed
130
131
132
133
134
| 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
135
136
| 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
137
| rec_r50fpn_vd_none_srn.yml    | SRN | Resnet50_fpn_vd    | None    | rnn | srn |
Topdu's avatar
Topdu committed
138
| rec_mtb_nrtr.yml    | NRTR | nrtr_mtb    | None    | transformer encoder | transformer decoder |
andyjpaddle's avatar
andyjpaddle committed
139
| rec_r31_sar.yml               | SAR | ResNet31 | None | LSTM encoder | LSTM decoder |
Khanh Tran's avatar
Khanh Tran committed
140
141


WenmuZhou's avatar
WenmuZhou committed
142
For training Chinese data, it is recommended to use
xmy0916's avatar
xmy0916 committed
143
[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
144
co
xmy0916's avatar
xmy0916 committed
145
Take `rec_chinese_lite_train_v2.0.yml` as an example:
Khanh Tran's avatar
Khanh Tran committed
146
147
148
```
Global:
  ...
xmy0916's avatar
xmy0916 committed
149
150
  # 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
151
152
  # Modify character type
  ...
153
  # Whether to recognize spaces
xmy0916's avatar
xmy0916 committed
154
  use_space_char: True
Khanh Tran's avatar
Khanh Tran committed
155

156
157
158
159

Optimizer:
  ...
  # Add learning rate decay strategy
xmy0916's avatar
xmy0916 committed
160
161
162
163
164
165
166
167
168
  lr:
    name: Cosine
    learning_rate: 0.001
  ...

...

Train:
  dataset:
MissPenguin's avatar
MissPenguin committed
169
    # Type of dataset,we support LMDBDataSet and SimpleDataSet
xmy0916's avatar
xmy0916 committed
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
    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
189
    # Type of dataset,we support LMDBDataSet and SimpleDataSet
xmy0916's avatar
xmy0916 committed
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
    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
205
206
207
```
**Note that the configuration file for prediction/evaluation must be consistent with the training.**

WenmuZhou's avatar
WenmuZhou committed
208
<a name="Multi_language"></a>
tink2123's avatar
tink2123 committed
209
### 2.3 Multi-language Training
tink2123's avatar
tink2123 committed
210
211
212

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

tink2123's avatar
tink2123 committed
213
214
215
216
217
218
219
220
221
222
223
224
| Configuration file |  Algorithm name |   backbone |   trans   |   seq      |     pred     |  language |
| :--------: |  :-------:   | :-------:  |   :-------:   |   :-----:   |  :-----:   | :-----:  |
| rec_chinese_cht_lite_train.yml |  CRNN |   Mobilenet_v3 small 0.5 |  None   |  BiLSTM |  ctc  | chinese traditional  |
| rec_en_lite_train.yml |  CRNN |   Mobilenet_v3 small 0.5 |  None   |  BiLSTM |  ctc  | English(Case sensitive)   |
| rec_french_lite_train.yml |  CRNN |   Mobilenet_v3 small 0.5 |  None   |  BiLSTM |  ctc  | French |  
| rec_ger_lite_train.yml |  CRNN |   Mobilenet_v3 small 0.5 |  None   |  BiLSTM |  ctc  | German   |
| rec_japan_lite_train.yml |  CRNN |   Mobilenet_v3 small 0.5 |  None   |  BiLSTM |  ctc  | Japanese |
| rec_korean_lite_train.yml |  CRNN |   Mobilenet_v3 small 0.5 |  None   |  BiLSTM |  ctc  | Korean  |
| rec_latin_lite_train.yml |  CRNN |   Mobilenet_v3 small 0.5 |  None   |  BiLSTM |  ctc  | Latin  |
| rec_arabic_lite_train.yml |  CRNN |   Mobilenet_v3 small 0.5 |  None   |  BiLSTM |  ctc  | arabic |
| rec_cyrillic_lite_train.yml |  CRNN |   Mobilenet_v3 small 0.5 |  None   |  BiLSTM |  ctc  | cyrillic   |
| rec_devanagari_lite_train.yml |  CRNN |   Mobilenet_v3 small 0.5 |  None   |  BiLSTM |  ctc  | devanagari  |
tink2123's avatar
tink2123 committed
225

tink2123's avatar
tink2123 committed
226
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
227
228
229
230
231
232
233
234
235


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
236
  # Add a custom dictionary, such as modify the dictionary, please point the path to the new dictionary
WenmuZhou's avatar
WenmuZhou committed
237
238
  character_dict_path: ./ppocr/utils/dict/french_dict.txt
  ...
xmy0916's avatar
xmy0916 committed
239
  # Whether to recognize spaces
xmy0916's avatar
xmy0916 committed
240
  use_space_char: True
xmy0916's avatar
xmy0916 committed
241

WenmuZhou's avatar
WenmuZhou committed
242
...
xmy0916's avatar
xmy0916 committed
243
244
245

Train:
  dataset:
MissPenguin's avatar
MissPenguin committed
246
    # Type of dataset,we support LMDBDataSet and SimpleDataSet
xmy0916's avatar
xmy0916 committed
247
248
249
250
251
252
253
254
255
    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
256
    # Type of dataset,we support LMDBDataSet and SimpleDataSet
xmy0916's avatar
xmy0916 committed
257
258
259
260
261
262
    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
263
```
Khanh Tran's avatar
Khanh Tran committed
264

265
266
267
268
269
270
<a name="kd"></a>

### 2.4 Training with Knowledge Distillation

Knowledge distillation is supported in PaddleOCR for text recognition training process. For more details, please refer to [doc](./knowledge_distillation_en.md).

WenmuZhou's avatar
WenmuZhou committed
271
<a name="EVALUATION"></a>
272

273
## 3. Evalution
Khanh Tran's avatar
Khanh Tran committed
274

WenmuZhou's avatar
WenmuZhou committed
275
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
276
277
278

```
# GPU evaluation, Global.checkpoints is the weight to be tested
WenmuZhou's avatar
WenmuZhou committed
279
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
280
281
```

WenmuZhou's avatar
WenmuZhou committed
282
<a name="PREDICTION"></a>
283
## 4. Prediction
Khanh Tran's avatar
Khanh Tran committed
284
285
286
287


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

tink2123's avatar
tink2123 committed
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
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
309
310
311

```
# Predict English results
WenmuZhou's avatar
WenmuZhou committed
312
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
313
314
```

tink2123's avatar
tink2123 committed
315

Khanh Tran's avatar
Khanh Tran committed
316
317
Input image:

318
![](../imgs_words/en/word_1.png)
Khanh Tran's avatar
Khanh Tran committed
319
320
321
322
323

Get the prediction result of the input image:

```
infer_img: doc/imgs_words/en/word_1.png
tink2123's avatar
tink2123 committed
324
        result: ('joint', 0.9998967)
Khanh Tran's avatar
Khanh Tran committed
325
326
```

xmy0916's avatar
xmy0916 committed
327
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
328
329
330

```
# Predict Chinese results
WenmuZhou's avatar
WenmuZhou committed
331
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
332
333
334
335
```

Input image:

336
![](../imgs_words/ch/word_1.jpg)
Khanh Tran's avatar
Khanh Tran committed
337
338
339
340
341

Get the prediction result of the input image:

```
infer_img: doc/imgs_words/ch/word_1.jpg
tink2123's avatar
tink2123 committed
342
        result: ('韩国小馆', 0.997218)
Khanh Tran's avatar
Khanh Tran committed
343
```
344
345
346

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

347
## 5. Convert to Inference Model
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

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

WenmuZhou's avatar
WenmuZhou committed
373
  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`
374
375

  ```
WenmuZhou's avatar
WenmuZhou committed
376
  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_dict_path="your text dict path"
377
  ```