未验证 提交 94cce909 编写于 作者: M MissPenguin 提交者: GitHub

Merge pull request #1339 from WenmuZhou/py_inference_doc

[Dygraph] add py inference doc
......@@ -8,7 +8,6 @@ Global:
# evaluation is run every 5000 iterations after the 4000th iteration
eval_batch_step: [0, 1000]
# if pretrained_model is saved in static mode, load_static_weights must set to True
load_static_weights: True
cal_metric_during_train: True
pretrained_model:
checkpoints:
......
......@@ -81,7 +81,8 @@ cv::Mat Classifier::Run(cv::Mat &img) {
void Classifier::LoadModel(const std::string &model_dir) {
AnalysisConfig config;
config.SetModel(model_dir + ".pdmodel", model_dir + ".pdiparams");
config.SetModel(model_dir + "/inference.pdmodel",
model_dir + "/inference.pdiparams");
if (this->use_gpu_) {
config.EnableUseGpu(this->gpu_mem_, this->gpu_id_);
......
......@@ -18,7 +18,8 @@ namespace PaddleOCR {
void DBDetector::LoadModel(const std::string &model_dir) {
AnalysisConfig config;
config.SetModel(model_dir + ".pdmodel", model_dir + ".pdiparams");
config.SetModel(model_dir + "/inference.pdmodel",
model_dir + "/inference.pdiparams");
if (this->use_gpu_) {
config.EnableUseGpu(this->gpu_mem_, this->gpu_id_);
......
......@@ -103,7 +103,8 @@ void CRNNRecognizer::Run(std::vector<std::vector<std::vector<int>>> boxes,
void CRNNRecognizer::LoadModel(const std::string &model_dir) {
AnalysisConfig config;
config.SetModel(model_dir + ".pdmodel", model_dir + ".pdiparams");
config.SetModel(model_dir + "/inference.pdmodel",
model_dir + "/inference.pdiparams");
if (this->use_gpu_) {
config.EnableUseGpu(this->gpu_mem_, this->gpu_id_);
......@@ -186,4 +187,4 @@ cv::Mat CRNNRecognizer::GetRotateCropImage(const cv::Mat &srcimage,
}
}
} // namespace PaddleOCR
} // namespace PaddleOCR
\ No newline at end of file
......@@ -62,9 +62,9 @@ PaddleOCR提供了训练脚本、评估脚本和预测脚本。
*如果您安装的是cpu版本,请将配置文件中的 `use_gpu` 字段修改为false*
```
# GPU训练 支持单卡,多卡训练,通过selected_gpus指定卡号
# GPU训练 支持单卡,多卡训练,通过gpus指定卡号
# 启动训练,下面的命令已经写入train.sh文件中,只需修改文件里的配置文件路径即可
python3 -m paddle.distributed.launch --selected_gpus '0,1,2,3,4,5,6,7' tools/train.py -c configs/cls/cls_mv3.yml
python3 -m paddle.distributed.launch --gpus '0,1,2,3,4,5,6,7' tools/train.py -c configs/cls/cls_mv3.yml
```
- 数据增强
......
# 基于Python预测引擎推理
inference 模型(`fluid.io.save_inference_model`保存的模型)
inference 模型(`paddle.jit.save`保存的模型)
一般是模型训练完成后保存的固化模型,多用于预测部署。训练过程中保存的模型是checkpoints模型,保存的是模型的参数,多用于恢复训练等。
与checkpoints模型相比,inference 模型会额外保存模型的结构信息,在预测部署、加速推理上性能优越,灵活方便,适合与实际系统集成。更详细的介绍请参考文档[分类预测框架](https://github.com/PaddlePaddle/PaddleClas/blob/master/docs/zh_CN/extension/paddle_inference.md).
与checkpoints模型相比,inference 模型会额外保存模型的结构信息,在预测部署、加速推理上性能优越,灵活方便,适合与实际系统集成。
接下来首先介绍如何将训练的模型转换成inference模型,然后将依次介绍文本检测、文本识别以及两者串联基于预测引擎推理。
接下来首先介绍如何将训练的模型转换成inference模型,然后将依次介绍文本检测、文本角度分类器、文本识别以及三者串联基于预测引擎推理。
- [一、训练模型转inference模型](#训练模型转inference模型)
......@@ -23,9 +23,8 @@ inference 模型(`fluid.io.save_inference_model`保存的模型)
- [1. 超轻量中文识别模型推理](#超轻量中文识别模型推理)
- [2. 基于CTC损失的识别模型推理](#基于CTC损失的识别模型推理)
- [3. 基于Attention损失的识别模型推理](#基于Attention损失的识别模型推理)
- [4. 基于SRN损失的识别模型推理](#基于SRN损失的识别模型推理)
- [5. 自定义文本识别字典的推理](#自定义文本识别字典的推理)
- [6. 多语言模型的推理](#多语言模型的推理)
- [4. 自定义文本识别字典的推理](#自定义文本识别字典的推理)
- [5. 多语言模型的推理](#多语言模型的推理)
- [四、方向分类模型推理](#方向识别模型推理)
- [1. 方向分类模型推理](#方向分类模型推理)
......@@ -42,24 +41,25 @@ inference 模型(`fluid.io.save_inference_model`保存的模型)
下载超轻量级中文检测模型:
```
wget -P ./ch_lite/ https://paddleocr.bj.bcebos.com/20-09-22/mobile/det/ch_ppocr_mobile_v1.1_det_train.tar && tar xf ./ch_lite/ch_ppocr_mobile_v1.1_det_train.tar -C ./ch_lite/
wget -P ./ch_lite/ {link} && tar xf ./ch_lite/ch_ppocr_mobile_v2.0_det_train.tar -C ./ch_lite/
```
上述模型是以MobileNetV3为backbone训练的DB算法,将训练好的模型转换成inference模型只需要运行如下命令:
```
# -c后面设置训练算法的yml配置文件
# -o配置可选参数
# Global.checkpoints参数设置待转换的训练模型地址,不用添加文件后缀.pdmodel,.pdopt或.pdparams。
# -c 后面设置训练算法的yml配置文件
# -o 配置可选参数
# Global.pretrained_model 参数设置待转换的训练模型地址,不用添加文件后缀 .pdmodel,.pdopt或.pdparams。
# Global.load_static_weights 参数需要设置为 False。
# Global.save_inference_dir参数设置转换的模型将保存的地址。
python3 tools/export_model.py -c configs/det/det_mv3_db_v1.1.yml -o Global.checkpoints=./ch_lite/ch_ppocr_mobile_v1.1_det_train/best_accuracy Global.save_inference_dir=./inference/det_db/
python3 tools/export_model.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o Global.checkpoints=./ch_lite/ch_ppocr_mobile_v2.0_det_train/best_accuracy Global.load_static_weights=False Global.save_inference_dir=./inference/det_db/
```
转inference模型时,使用的配置文件和训练时使用的配置文件相同。另外,还需要设置配置文件中的`Global.checkpoints``Global.save_inference_dir`参数。
其中`Global.checkpoints`指向训练中保存的模型参数文件,`Global.save_inference_dir`是生成的inference模型要保存的目录。
转换成功后,在`save_inference_dir`目录下有两个文件:
转inference模型时,使用的配置文件和训练时使用的配置文件相同。另外,还需要设置配置文件中的`Global.checkpoints`参数,其指向训练中保存的模型参数文件。
转换成功后,在模型保存目录下有三个文件:
```
inference/det_db/
└─ model 检测inference模型的program文件
└─ params 检测inference模型的参数文件
├── inference.pdiparams # 检测inference模型的参数文件
├── inference.pdiparams.info # 检测inference模型的参数信息,可忽略
└── inference.pdmodel # 检测inference模型的program文件
```
<a name="识别模型转inference模型"></a>
......@@ -67,27 +67,28 @@ inference/det_db/
下载超轻量中文识别模型:
```
wget -P ./ch_lite/ https://paddleocr.bj.bcebos.com/20-09-22/mobile/rec/ch_ppocr_mobile_v1.1_rec_train.tar && tar xf ./ch_lite/ch_ppocr_mobile_v1.1_rec_train.tar -C ./ch_lite/
wget -P ./ch_lite/ {link} && tar xf ./ch_lite/ch_ppocr_mobile_v2.0_rec_train.tar -C ./ch_lite/
```
识别模型转inference模型与检测的方式相同,如下:
```
# -c后面设置训练算法的yml配置文件
# -o配置可选参数
# Global.checkpoints参数设置待转换的训练模型地址,不用添加文件后缀.pdmodel,.pdopt或.pdparams。
# -c 后面设置训练算法的yml配置文件
# -o 配置可选参数
# Global.pretrained_model 参数设置待转换的训练模型地址,不用添加文件后缀 .pdmodel,.pdopt或.pdparams。
# Global.load_static_weights 参数需要设置为 False。
# Global.save_inference_dir参数设置转换的模型将保存的地址。
python3 tools/export_model.py -c configs/rec/ch_ppocr_v1.1/rec_chinese_lite_train_v1.1.yml -o Global.checkpoints=./ch_lite/ch_ppocr_mobile_v1.1_rec_train/best_accuracy \
Global.save_inference_dir=./inference/rec_crnn/
python3 tools/export_model.py -c configs/rec/ch_ppocr_v2.0/rec_chinese_lite_train_v2.0.yml -o Global.checkpoints=./ch_lite/ch_ppocr_mobile_v2.0_rec_train/best_accuracy Global.load_static_weights=False Global.save_inference_dir=./inference/rec_crnn/
```
**注意:**如果您是在自己的数据集上训练的模型,并且调整了中文字符的字典文件,请注意修改配置文件中的`character_dict_path`是否是所需要的字典文件。
转换成功后,在目录下有个文件:
转换成功后,在目录下有个文件:
```
/inference/rec_crnn/
└─ model 识别inference模型的program文件
└─ params 识别inference模型的参数文件
├── 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
```
<a name="方向分类模型转inference模型"></a>
......@@ -95,25 +96,26 @@ python3 tools/export_model.py -c configs/rec/ch_ppocr_v1.1/rec_chinese_lite_trai
下载方向分类模型:
```
wget -P ./ch_lite/ https://paddleocr.bj.bcebos.com/20-09-22/cls/ch_ppocr_mobile_v1.1_cls_train.tar && tar xf ./ch_lite/ch_ppocr_mobile_v1.1_cls_train.tar -C ./ch_lite/
wget -P ./ch_lite/ {link} && tar xf ./ch_lite/ch_ppocr_mobile_v2.0_cls_train.tar -C ./ch_lite/
```
方向分类模型转inference模型与检测的方式相同,如下:
```
# -c后面设置训练算法的yml配置文件
# -o配置可选参数
# Global.checkpoints参数设置待转换的训练模型地址,不用添加文件后缀.pdmodel,.pdopt或.pdparams。
# -c 后面设置训练算法的yml配置文件
# -o 配置可选参数
# Global.pretrained_model 参数设置待转换的训练模型地址,不用添加文件后缀 .pdmodel,.pdopt或.pdparams。
# Global.load_static_weights 参数需要设置为 False。
# Global.save_inference_dir参数设置转换的模型将保存的地址。
python3 tools/export_model.py -c configs/cls/cls_mv3.yml -o Global.checkpoints=./ch_lite/ch_ppocr_mobile_v1.1_cls_train/best_accuracy \
Global.save_inference_dir=./inference/cls/
python3 tools/export_model.py -c configs/cls/cls_mv3.yml -o Global.checkpoints=./ch_lite/ch_ppocr_mobile_v2.0_cls_train/best_accuracy Global.load_static_weights=False Global.save_inference_dir=./inference/cls/
```
转换成功后,在目录下有个文件:
转换成功后,在目录下有个文件:
```
/inference/cls/
└─ model 识别inference模型的program文件
└─ params 识别inference模型的参数文件
├── inference.pdiparams # 分类inference模型的参数文件
├── inference.pdiparams.info # 分类inference模型的参数信息,可忽略
└── inference.pdmodel # 分类inference模型的program文件
```
<a name="文本检测模型推理"></a>
......@@ -134,10 +136,12 @@ python3 tools/infer/predict_det.py --image_dir="./doc/imgs/2.jpg" --det_model_di
![](../imgs_results/det_res_2.jpg)
通过设置参数`det_max_side_len`的大小,改变检测算法中图片规范化的最大值。当图片的长宽都小于`det_max_side_len`,则使用原图预测,否则将图片等比例缩放到最大值,进行预测。该参数默认设置为`det_max_side_len=960`。 如果输入图片的分辨率比较大,而且想使用更大的分辨率预测,可以执行如下命令:
通过参数`limit_type``det_limit_side_len`来对图片的尺寸进行限制限,`limit_type=max`为限制长边长度<`det_limit_side_len`,`limit_type=min`为限制短边长度>`det_limit_side_len`,
图片不满足限制条件时(`limit_type=max`时长边长度>`det_limit_side_len``limit_type=min`时短边长度<`det_limit_side_len`),将对图片进行等比例缩放。
该参数默认设置为`limit_type='max',det_max_side_len=960`。 如果输入图片的分辨率比较大,而且想使用更大的分辨率预测,可以执行如下命令:
```
python3 tools/infer/predict_det.py --image_dir="./doc/imgs/2.jpg" --det_model_dir="./inference/det_db/" --det_max_side_len=1200
python3 tools/infer/predict_det.py --image_dir="./doc/imgs/2.jpg" --det_model_dir="./inference/det_db/" --det_limit_type=max --det_limit_side_len=1200
```
如果想使用CPU进行预测,执行命令如下
......@@ -148,14 +152,10 @@ python3 tools/infer/predict_det.py --image_dir="./doc/imgs/2.jpg" --det_model_di
<a name="DB文本检测模型推理"></a>
### 2. DB文本检测模型推理
首先将DB文本检测训练过程中保存的模型,转换成inference model。以基于Resnet50_vd骨干网络,在ICDAR2015英文数据集训练的模型为例([模型下载地址](https://paddleocr.bj.bcebos.com/det_r50_vd_db.tar)),可以使用如下命令进行转换:
首先将DB文本检测训练过程中保存的模型,转换成inference model。以基于Resnet50_vd骨干网络,在ICDAR2015英文数据集训练的模型为例([模型下载地址](link)),可以使用如下命令进行转换:
```
# -c后面设置训练算法的yml配置文件
# Global.checkpoints参数设置待转换的训练模型地址,不用添加文件后缀.pdmodel,.pdopt或.pdparams。
# Global.save_inference_dir参数设置转换的模型将保存的地址。
python3 tools/export_model.py -c configs/det/det_r50_vd_db.yml -o Global.checkpoints="./models/det_r50_vd_db/best_accuracy" Global.save_inference_dir="./inference/det_db"
python3 tools/export_model.py -c configs/det/det_r50_vd_db.yml -o Global.checkpoints=./det_r50_vd_db_v2.0.train/best_accuracy Global.load_static_weights=False Global.save_inference_dir=./inference/det_db
```
DB文本检测模型推理,可以执行如下命令:
......@@ -173,14 +173,10 @@ python3 tools/infer/predict_det.py --image_dir="./doc/imgs_en/img_10.jpg" --det_
<a name="EAST文本检测模型推理"></a>
### 3. EAST文本检测模型推理
首先将EAST文本检测训练过程中保存的模型,转换成inference model。以基于Resnet50_vd骨干网络,在ICDAR2015英文数据集训练的模型为例([模型下载地址](https://paddleocr.bj.bcebos.com/det_r50_vd_east.tar)),可以使用如下命令进行转换:
首先将EAST文本检测训练过程中保存的模型,转换成inference model。以基于Resnet50_vd骨干网络,在ICDAR2015英文数据集训练的模型为例([模型下载地址](link)),可以使用如下命令进行转换:
```
# -c后面设置训练算法的yml配置文件
# Global.checkpoints参数设置待转换的训练模型地址,不用添加文件后缀.pdmodel,.pdopt或.pdparams。
# Global.save_inference_dir参数设置转换的模型将保存的地址。
python3 tools/export_model.py -c configs/det/det_r50_vd_east.yml -o Global.checkpoints="./models/det_r50_vd_east/best_accuracy" Global.save_inference_dir="./inference/det_east"
python3 tools/export_model.py -c configs/det/det_r50_vd_east.yml -o Global.checkpoints=./det_r50_vd_east_v2.0.train/best_accuracy Global.load_static_weights=False Global.save_inference_dir=./inference/det_east
```
**EAST文本检测模型推理,需要设置参数`--det_algorithm="EAST"`**,可以执行如下命令:
......@@ -198,9 +194,10 @@ python3 tools/infer/predict_det.py --det_algorithm="EAST" --image_dir="./doc/img
<a name="SAST文本检测模型推理"></a>
### 4. SAST文本检测模型推理
#### (1). 四边形文本检测模型(ICDAR2015)
首先将SAST文本检测训练过程中保存的模型,转换成inference model。以基于Resnet50_vd骨干网络,在ICDAR2015英文数据集训练的模型为例([模型下载地址](https://paddleocr.bj.bcebos.com/SAST/sast_r50_vd_icdar2015.tar)),可以使用如下命令进行转换:
首先将SAST文本检测训练过程中保存的模型,转换成inference model。以基于Resnet50_vd骨干网络,在ICDAR2015英文数据集训练的模型为例([模型下载地址](link)),可以使用如下命令进行转换:
```
python3 tools/export_model.py -c configs/det/det_r50_vd_sast_icdar15.yml -o Global.checkpoints="./models/sast_r50_vd_icdar2015/best_accuracy" Global.save_inference_dir="./inference/det_sast_ic15"
python3 tools/export_model.py -c configs/det/det_r50_vd_sast_icdar15.yml -o Global.checkpoints=./det_r50_vd_sast_icdar15_v2.0.train/best_accuracy Global.load_static_weights=False Global.save_inference_dir=./inference/det_sast_ic15
```
**SAST文本检测模型推理,需要设置参数`--det_algorithm="SAST"`**,可以执行如下命令:
```
......@@ -211,10 +208,11 @@ python3 tools/infer/predict_det.py --det_algorithm="SAST" --image_dir="./doc/img
![](../imgs_results/det_res_img_10_sast.jpg)
#### (2). 弯曲文本检测模型(Total-Text)
首先将SAST文本检测训练过程中保存的模型,转换成inference model。以基于Resnet50_vd骨干网络,在Total-Text英文数据集训练的模型为例([模型下载地址](https://paddleocr.bj.bcebos.com/SAST/sast_r50_vd_total_text.tar)),可以使用如下命令进行转换:
首先将SAST文本检测训练过程中保存的模型,转换成inference model。以基于Resnet50_vd骨干网络,在Total-Text英文数据集训练的模型为例([模型下载地址](link)),可以使用如下命令进行转换:
```
python3 tools/export_model.py -c configs/det/det_r50_vd_sast_totaltext.yml -o Global.checkpoints="./models/sast_r50_vd_total_text/best_accuracy" Global.save_inference_dir="./inference/det_sast_tt"
python3 tools/export_model.py -c configs/det/det_r50_vd_sast_totaltext.yml -o Global.checkpoints=./det_r50_vd_sast_totaltext_v2.0.train/best_accuracy Global.load_static_weights=False Global.save_inference_dir=./inference/det_sast_tt
```
**SAST文本检测模型推理,需要设置参数`--det_algorithm="SAST"`,同时,还需要增加参数`--det_sast_polygon=True`,**可以执行如下命令:
......@@ -253,33 +251,30 @@ Predicts of ./doc/imgs_words/ch/word_4.jpg:['实力活力', 0.89552695]
<a name="基于CTC损失的识别模型推理"></a>
### 2. 基于CTC损失的识别模型推理
我们以STAR-Net为例,介绍基于CTC损失的识别模型推理。 CRNN和Rosetta使用方式类似,不用设置识别算法参数rec_algorithm。
我们以 CRNN 为例,介绍基于CTC损失的识别模型推理。 Rosetta 使用方式类似,不用设置识别算法参数rec_algorithm。
首先将STAR-Net文本识别训练过程中保存的模型,转换成inference model。以基于Resnet34_vd骨干网络,使用MJSynth和SynthText两个英文文本识别合成数据集训练
的模型为例([模型下载地址](https://paddleocr.bj.bcebos.com/rec_r34_vd_tps_bilstm_ctc.tar)),可以使用如下命令进行转换:
首先将 Rosetta 文本识别训练过程中保存的模型,转换成inference model。以基于Resnet34_vd骨干网络,使用MJSynth和SynthText两个英文文本识别合成数据集训练
的模型为例([模型下载地址](link)),可以使用如下命令进行转换:
```
# -c后面设置训练算法的yml配置文件
# Global.checkpoints参数设置待转换的训练模型地址,不用添加文件后缀.pdmodel,.pdopt或.pdparams。
# Global.save_inference_dir参数设置转换的模型将保存的地址。
python3 tools/export_model.py -c configs/det/rec_r34_vd_none_bilstm_ctc.yml -o Global.checkpoints=./rec_r34_vd_none_bilstm_ctc_v2.0.train/best_accuracy Global.load_static_weights=False Global.save_inference_dir=./inference/rec_crnn
python3 tools/export_model.py -c configs/rec/rec_r34_vd_tps_bilstm_ctc.yml -o Global.checkpoints="./models/rec_r34_vd_tps_bilstm_ctc/best_accuracy" Global.save_inference_dir="./inference/starnet"
```
STAR-Net文本识别模型推理,可以执行如下命令:
```
python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words_en/word_336.png" --rec_model_dir="./inference/starnet/" --rec_image_shape="3, 32, 100" --rec_char_type="en"
python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words_en/word_336.png" --rec_model_dir="./inference/rec_crnn/" --rec_image_shape="3, 32, 100" --rec_char_type="en"
```
<a name="基于Attention损失的识别模型推理"></a>
### 3. 基于Attention损失的识别模型推理
基于Attention损失的识别模型与ctc不同,需要额外设置识别算法参数 --rec_algorithm="RARE"
RARE 文本识别模型推理,可以执行如下命令:
```
python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words_en/word_336.png" --rec_model_dir="./inference/rare/" --rec_image_shape="3, 32, 100" --rec_char_type="en" --rec_algorithm="RARE"
```
![](../imgs_words_en/word_336.png)
......@@ -298,21 +293,8 @@ Predicts of ./doc/imgs_words_en/word_336.png:['super', 0.9999555]
self.character_str = "0123456789abcdefghijklmnopqrstuvwxyz"
dict_character = list(self.character_str)
```
<a name="基于SRN损失的识别模型推理"></a>
### 4. 基于SRN损失的识别模型推理
基于SRN损失的识别模型,需要额外设置识别算法参数 --rec_algorithm="SRN"。 同时需要保证预测shape与训练时一致,如: --rec_image_shape="1, 64, 256"
```
python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words_en/word_336.png" \
--rec_model_dir="./inference/srn/" \
--rec_image_shape="1, 64, 256" \
--rec_char_type="en" \
--rec_algorithm="SRN"
```
<a name="自定义文本识别字典的推理"></a>
### 5. 自定义文本识别字典的推理
### 4. 自定义文本识别字典的推理
如果训练时修改了文本的字典,在使用inference模型预测时,需要通过`--rec_char_dict_path`指定使用的字典路径
```
......@@ -320,7 +302,7 @@ python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words_en/word_336.png
```
<a name="多语言模型的推理"></a>
### 6. 多语言模型的推理
### 5. 多语言模型的推理
如果您需要预测的是其他语言模型,在使用inference模型预测时,需要通过`--rec_char_dict_path`指定使用的字典路径, 同时为了得到正确的可视化结果,
需要通过 `--vis_font_path` 指定可视化的字体路径,`doc/` 路径下有默认提供的小语种字体,例如韩文识别:
......@@ -350,11 +332,14 @@ python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words/korean/1.jpg" -
python3 tools/infer/predict_cls.py --image_dir="./doc/imgs_words/ch/word_4.jpg" --cls_model_dir="./inference/cls/"
```
![](../imgs_words/ch/word_4.jpg)
![](../imgs_words/ch/word_1.jpg)
执行命令后,上面图像的预测结果(分类的方向和得分)会打印到屏幕上,示例如下:
Predicts of ./doc/imgs_words/ch/word_4.jpg:['0', 0.9999963]
```
infer_img: doc/imgs_words/ch/word_1.jpg
result: ('0', 0.9998784)
```
<a name="文本检测、方向分类和文字识别串联推理"></a>
## 五、文本检测、方向分类和文字识别串联推理
......
......@@ -65,9 +65,9 @@ Start training:
```
# Set PYTHONPATH path
export PYTHONPATH=$PYTHONPATH:.
# GPU training Support single card and multi-card training, specify the card number through selected_gpus
# GPU training Support single card and multi-card training, specify the card number through gpus
# Start training, the following command has been written into the train.sh file, just modify the configuration file path in the file
python3 -m paddle.distributed.launch --selected_gpus '0,1,2,3,4,5,6,7' tools/train.py -c configs/cls/cls_mv3.yml
python3 -m paddle.distributed.launch --gpus '0,1,2,3,4,5,6,7' tools/train.py -c configs/cls/cls_mv3.yml
```
- Data Augmentation
......
# Reasoning based on Python prediction engine
The inference model (the model saved by `fluid.io.save_inference_model`) is generally a solidified model saved after the model training is completed, and is mostly used to give prediction in deployment.
The inference model (the model saved by `paddle.jit.save`) is generally a solidified model saved after the model training is completed, and is mostly used to give prediction in deployment.
The model saved during the training process is the checkpoints model, which saves the parameters of the model and is mostly used to resume training.
Compared with the checkpoints model, the inference model will additionally save the structural information of the model. It has superior performance in predicting in deployment and accelerating inferencing, is flexible and convenient, and is suitable for integration with actual systems. For more details, please refer to the document [Classification Framework](https://github.com/PaddlePaddle/PaddleClas/blob/master/docs/zh_CN/extension/paddle_inference.md).
Next, we first introduce how to convert a trained model into an inference model, and then we will introduce text detection, text recognition, and the concatenation of them based on inference model.
Next, we first introduce how to convert a trained model into an inference model, and then we will introduce text detection, text recognition, angle class, and the concatenation of them based on inference model.
- [CONVERT TRAINING MODEL TO INFERENCE MODEL](#CONVERT)
- [Convert detection model to inference model](#Convert_detection_model)
......@@ -26,9 +26,8 @@ Next, we first introduce how to convert a trained model into an inference model,
- [1. LIGHTWEIGHT CHINESE MODEL](#LIGHTWEIGHT_RECOGNITION)
- [2. CTC-BASED TEXT RECOGNITION MODEL INFERENCE](#CTC-BASED_RECOGNITION)
- [3. ATTENTION-BASED TEXT RECOGNITION MODEL INFERENCE](#ATTENTION-BASED_RECOGNITION)
- [4. SRN-BASED TEXT RECOGNITION MODEL INFERENCE](#SRN-BASED_RECOGNITION)
- [5. TEXT RECOGNITION MODEL INFERENCE USING CUSTOM CHARACTERS DICTIONARY](#USING_CUSTOM_CHARACTERS)
- [6. MULTILINGUAL MODEL INFERENCE](MULTILINGUAL_MODEL_INFERENCE)
- [4. TEXT RECOGNITION MODEL INFERENCE USING CUSTOM CHARACTERS DICTIONARY](#USING_CUSTOM_CHARACTERS)
- [5. MULTILINGUAL MODEL INFERENCE](MULTILINGUAL_MODEL_INFERENCE)
- [ANGLE CLASSIFICATION MODEL INFERENCE](#ANGLE_CLASS_MODEL_INFERENCE)
- [1. ANGLE CLASSIFICATION MODEL INFERENCE](#ANGLE_CLASS_MODEL_INFERENCE)
......@@ -44,26 +43,27 @@ Next, we first introduce how to convert a trained model into an inference model,
Download the lightweight Chinese detection model:
```
wget -P ./ch_lite/ https://paddleocr.bj.bcebos.com/20-09-22/mobile/det/ch_ppocr_mobile_v1.1_det_train.tar && tar xf ./ch_lite/ch_ppocr_mobile_v1.1_det_train.tar -C ./ch_lite/
wget -P ./ch_lite/ {link} && tar xf ./ch_lite/ch_ppocr_mobile_v2.0_det_train.tar -C ./ch_lite/
```
The above model is a DB algorithm trained with MobileNetV3 as the backbone. To convert the trained model into an inference model, just run the following command:
```
# -c Set the training algorithm yml configuration file
# -o Set optional parameters
# Global.checkpoints 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.
# Global.checkpoints parameter Set the training model address to be converted without adding the file suffix .pdmodel, .pdopt or .pdparams.
# Global.load_static_weights needs to be set to False
# Global.save_inference_dir Set the address where the converted model will be saved.
python3 tools/export_model.py -c configs/det/det_mv3_db_v1.1.yml -o Global.checkpoints=./ch_lite/ch_ppocr_mobile_v1.1_det_train/best_accuracy Global.save_inference_dir=./inference/det_db/
python3 tools/export_model.py -c configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml -o Global.checkpoints=./ch_lite/ch_ppocr_mobile_v2.0_det_train/best_accuracy Global.load_static_weights=False Global.save_inference_dir=./inference/det_db/
```
When converting to an inference model, the configuration file used is the same as the configuration file used during training. In addition, you also need to set the `Global.checkpoints` and `Global.save_inference_dir` parameters in the configuration file.
`Global.checkpoints` points to the model parameter file saved during training, and `Global.save_inference_dir` is the directory where the generated inference model is saved.
After the conversion is successful, there are two files in the `save_inference_dir` directory:
When converting to an inference model, the configuration file used is the same as the configuration file used during training. In addition, you also need to set the `Global.checkpoints` parameter in the configuration file.
After the conversion is successful, there are three files in the model save directory:
```
inference/det_db/
└─ model Check the program file of inference model
└─ params Check the parameter file of the inference model
├── inference.pdiparams # The parameter file of detection inference model
├── inference.pdiparams.info # The parameter information of detection inference model, which can be ignored
└── inference.pdmodel # The program file of detection inference model
```
<a name="Convert_recognition_model"></a>
......@@ -71,26 +71,28 @@ inference/det_db/
Download the lightweight Chinese recognition model:
```
wget -P ./ch_lite/ https://paddleocr.bj.bcebos.com/20-09-22/mobile/rec/ch_ppocr_mobile_v1.1_rec_train.tar && tar xf ch_ppocr_mobile_v1.1_rec_train.tar -C ./ch_lite/
wget -P ./ch_lite/ {link} && tar xf ./ch_lite/ch_ppocr_mobile_v2.0_rec_train.tar -C ./ch_lite/
```
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.checkpoints 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.
# Global.checkpoints parameter Set the training model address to be converted without adding the file suffix .pdmodel, .pdopt or .pdparams.
# Global.load_static_weights needs to be set to False
# Global.save_inference_dir Set the address where the converted model will be saved.
python3 tools/export_model.py -c configs/rec/ch_ppocr_v1.1/rec_chinese_lite_train_v1.1.yml -o Global.checkpoints=./ch_lite/ch_ppocr_mobile_v1.1_rec_train/best_accuracy \
python3 tools/export_model.py -c configs/rec/ch_ppocr_v2.0/rec_chinese_lite_train_v2.0.yml -o Global.checkpoints=./ch_lite/ch_ppocr_mobile_v2.0_rec_train/best_accuracy Global.load_static_weights=False 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 two files in the directory:
After the conversion is successful, there are three files in the model save directory:
```
/inference/rec_crnn/
└─ model Identify the saved model files
└─ params Identify the parameter files of the inference model
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
```
<a name="Convert_angle_class_model"></a>
......@@ -98,25 +100,26 @@ After the conversion is successful, there are two files in the directory:
Download the angle classification model:
```
wget -P ./ch_lite/ https://paddleocr.bj.bcebos.com/20-09-22/cls/ch_ppocr_mobile_v1.1_cls_train.tar && tar xf ./ch_lite/ch_ppocr_mobile_v1.1_cls_train.tar -C ./ch_lite/
wget -P ./ch_lite/ {link} && tar xf ./ch_lite/ch_ppocr_mobile_v2.0_cls_train.tar -C ./ch_lite/
```
The angle classification 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.checkpoints 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.
# Global.checkpoints parameter Set the training model address to be converted without adding the file suffix .pdmodel, .pdopt or .pdparams.
# Global.load_static_weights needs to be set to False
# Global.save_inference_dir Set the address where the converted model will be saved.
python3 tools/export_model.py -c configs/cls/cls_mv3.yml -o Global.checkpoints=./ch_lite/ch_ppocr_mobile_v1.1_cls_train/best_accuracy \
Global.save_inference_dir=./inference/cls/
python3 tools/export_model.py -c configs/cls/cls_mv3.yml -o Global.checkpoints=./ch_lite/ch_ppocr_mobile_v2.0_cls_train/best_accuracy Global.load_static_weights=False Global.save_inference_dir=./inference/cls/
```
After the conversion is successful, there are two files in the directory:
```
/inference/cls/
└─ model Identify the saved model files
└─ params Identify the parameter files of the inference model
inference/det_db/
├── inference.pdiparams # The parameter file of angle class inference model
├── inference.pdiparams.info # The parameter information of angle class inference model, which can be ignored
└── inference.pdmodel # The program file of angle class model
```
......@@ -139,10 +142,12 @@ The visual text detection results are saved to the ./inference_results folder by
![](../imgs_results/det_res_2.jpg)
By setting the size of the parameter `det_max_side_len`, the maximum value of picture normalization in the detection algorithm is changed. When the length and width of the picture are less than det_max_side_len, the original picture is used for prediction, otherwise the picture is scaled to the maximum value for prediction. This parameter is set to det_max_side_len=960 by default. If the resolution of the input picture is relatively large and you want to use a larger resolution for prediction, you can execute the following command:
The size of the image is limited by the parameters `limit_type` and `det_limit_side_len`, `limit_type=max` is to limit the length of the long side <`det_limit_side_len`, and `limit_type=min` is to limit the length of the short side>`det_limit_side_len`,
When the picture does not meet the restriction conditions (for `limit_type=max`and long side >`det_limit_side_len` or for `min` and short side <`det_limit_side_len`), the image will be scaled proportionally.
This parameter is set to `limit_type='max', det_max_side_len=960` by default. If the resolution of the input picture is relatively large, and you want to use a larger resolution prediction, you can execute the following command:
```
python3 tools/infer/predict_det.py --image_dir="./doc/imgs/2.jpg" --det_model_dir="./inference/det_db/" --det_max_side_len=1200
python3 tools/infer/predict_det.py --image_dir="./doc/imgs/2.jpg" --det_model_dir="./inference/det_db/" --det_limit_type=max --det_limit_side_len=1200
```
If you want to use the CPU for prediction, execute the command as follows
......@@ -153,14 +158,10 @@ python3 tools/infer/predict_det.py --image_dir="./doc/imgs/2.jpg" --det_model_di
<a name="DB_DETECTION"></a>
### 2. DB TEXT DETECTION MODEL INFERENCE
First, convert the model saved in the DB text detection training process into an inference model. Taking the model based on the Resnet50_vd backbone network and trained on the ICDAR2015 English dataset as an example ([model download link](https://paddleocr.bj.bcebos.com/det_r50_vd_db.tar)), you can use the following command to convert:
First, convert the model saved in the DB text detection training process into an inference model. Taking the model based on the Resnet50_vd backbone network and trained on the ICDAR2015 English dataset as an example ([model download link](link)), you can use the following command to convert:
```
# Set the yml configuration file of the training algorithm after -c
# The Global.checkpoints parameter sets the address of the training model to be converted without adding the file suffix .pdmodel, .pdopt or .pdparams.
# The Global.save_inference_dir parameter sets the address where the converted model will be saved.
python3 tools/export_model.py -c configs/det/det_r50_vd_db.yml -o Global.checkpoints="./models/det_r50_vd_db/best_accuracy" Global.save_inference_dir="./inference/det_db"
python3 tools/export_model.py -c configs/det/det_r50_vd_db.yml -o Global.checkpoints=./det_r50_vd_db_v2.0.train/best_accuracy Global.load_static_weights=False Global.save_inference_dir=./inference/det_db
```
DB text detection model inference, you can execute the following command:
......@@ -178,16 +179,11 @@ The visualized text detection results are saved to the `./inference_results` fol
<a name="EAST_DETECTION"></a>
### 3. EAST TEXT DETECTION MODEL INFERENCE
First, convert the model saved in the EAST text detection training process into an inference model. Taking the model based on the Resnet50_vd backbone network and trained on the ICDAR2015 English dataset as an example ([model download link](https://paddleocr.bj.bcebos.com/det_r50_vd_east.tar)), you can use the following command to convert:
First, convert the model saved in the EAST text detection training process into an inference model. Taking the model based on the Resnet50_vd backbone network and trained on the ICDAR2015 English dataset as an example ([model download link](link)), you can use the following command to convert:
```
# Set the yml configuration file of the training algorithm after -c
# The Global.checkpoints parameter sets the address of the training model to be converted without adding the file suffix .pdmodel, .pdopt or .pdparams.
# The Global.save_inference_dir parameter sets the address where the converted model will be saved.
python3 tools/export_model.py -c configs/det/det_r50_vd_east.yml -o Global.checkpoints="./models/det_r50_vd_east/best_accuracy" Global.save_inference_dir="./inference/det_east"
python3 tools/export_model.py -c configs/det/det_r50_vd_east.yml -o Global.checkpoints=./det_r50_vd_east_v2.0.train/best_accuracy Global.load_static_weights=False Global.save_inference_dir=./inference/det_east
```
**For EAST text detection model inference, you need to set the parameter ``--det_algorithm="EAST"``**, run the following command:
```
......@@ -204,10 +200,10 @@ The visualized text detection results are saved to the `./inference_results` fol
<a name="SAST_DETECTION"></a>
### 4. SAST TEXT DETECTION MODEL INFERENCE
#### (1). Quadrangle text detection model (ICDAR2015)
First, convert the model saved in the SAST text detection training process into an inference model. Taking the model based on the Resnet50_vd backbone network and trained on the ICDAR2015 English dataset as an example ([model download link](https://paddleocr.bj.bcebos.com/SAST/sast_r50_vd_icdar2015.tar)), you can use the following command to convert:
First, convert the model saved in the SAST text detection training process into an inference model. Taking the model based on the Resnet50_vd backbone network and trained on the ICDAR2015 English dataset as an example ([model download link](link)), you can use the following command to convert:
```
python3 tools/export_model.py -c configs/det/det_r50_vd_sast_icdar15.yml -o Global.checkpoints="./models/sast_r50_vd_icdar2015/best_accuracy" Global.save_inference_dir="./inference/det_sast_ic15"
python3 tools/export_model.py -c configs/det/det_r50_vd_sast_icdar15.yml -o Global.checkpoints=./det_r50_vd_sast_icdar15_v2.0.train/best_accuracy Global.load_static_weights=False Global.save_inference_dir=./inference/det_sast_ic15
```
**For SAST quadrangle text detection model inference, you need to set the parameter `--det_algorithm="SAST"`**, run the following command:
......@@ -224,7 +220,7 @@ The visualized text detection results are saved to the `./inference_results` fol
First, convert the model saved in the SAST text detection training process into an inference model. Taking the model based on the Resnet50_vd backbone network and trained on the Total-Text English dataset as an example ([model download link](https://paddleocr.bj.bcebos.com/SAST/sast_r50_vd_total_text.tar)), you can use the following command to convert:
```
python3 tools/export_model.py -c configs/det/det_r50_vd_sast_totaltext.yml -o Global.checkpoints="./models/sast_r50_vd_total_text/best_accuracy" Global.save_inference_dir="./inference/det_sast_tt"
python3 tools/export_model.py -c configs/det/det_r50_vd_sast_totaltext.yml -o Global.checkpoints=./det_r50_vd_sast_totaltext_v2.0.train/best_accuracy Global.load_static_weights=False Global.save_inference_dir=./inference/det_sast_tt
```
**For SAST curved text detection model inference, you need to set the parameter `--det_algorithm="SAST"` and `--det_sast_polygon=True`**, run the following command:
......@@ -264,19 +260,15 @@ Predicts of ./doc/imgs_words/ch/word_4.jpg:['实力活力', 0.89552695]
<a name="CTC-BASED_RECOGNITION"></a>
### 2. CTC-BASED TEXT RECOGNITION MODEL INFERENCE
Taking STAR-Net as an example, we introduce the recognition model inference based on CTC loss. CRNN and Rosetta are used in a similar way, by setting the recognition algorithm parameter `rec_algorithm`.
Taking CRNN as an example, we introduce the recognition model inference based on CTC loss. Rosetta and Star-Net are used in a similar way, No need to set the recognition algorithm parameter rec_algorithm.
First, convert the model saved in the STAR-Net text recognition training process into an inference model. Taking the model based on Resnet34_vd backbone network, using MJSynth and SynthText (two English text recognition synthetic datasets) for training, as an example ([model download address](https://paddleocr.bj.bcebos.com/rec_r34_vd_tps_bilstm_ctc.tar)). It can be converted as follow:
First, convert the model saved in the CRNN text recognition training process into an inference model. Taking the model based on Resnet34_vd backbone network, using MJSynth and SynthText (two English text recognition synthetic datasets) for training, as an example ([model download address](link)). It can be converted as follow:
```
# Set the yml configuration file of the training algorithm after -c
# The Global.checkpoints parameter sets the address of the training model to be converted without adding the file suffix .pdmodel, .pdopt or .pdparams.
# The Global.save_inference_dir parameter sets the address where the converted model will be saved.
python3 tools/export_model.py -c configs/rec/rec_r34_vd_tps_bilstm_ctc.yml -o Global.checkpoints="./models/rec_r34_vd_tps_bilstm_ctc/best_accuracy" Global.save_inference_dir="./inference/starnet"
python3 tools/export_model.py -c configs/det/rec_r34_vd_none_bilstm_ctc.yml -o Global.checkpoints=./rec_r34_vd_none_bilstm_ctc_v2.0.train/best_accuracy Global.load_static_weights=False Global.save_inference_dir=./inference/rec_crnn
```
For STAR-Net text recognition model inference, execute the following commands:
For CRNN text recognition model inference, execute the following commands:
```
python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words_en/word_336.png" --rec_model_dir="./inference/starnet/" --rec_image_shape="3, 32, 100" --rec_char_type="en"
......@@ -286,7 +278,11 @@ python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words_en/word_336.png
### 3. ATTENTION-BASED TEXT RECOGNITION MODEL INFERENCE
![](../imgs_words_en/word_336.png)
The recognition model based on Attention loss is different from ctc, and additional recognition algorithm parameters need to be set --rec_algorithm="RARE"
After executing the command, the recognition result of the above image is as follows:
```bash
python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words_en/word_336.png" --rec_model_dir="./inference/rare/" --rec_image_shape="3, 32, 100" --rec_char_type="en" --rec_algorithm="RARE"
```
Predicts of ./doc/imgs_words_en/word_336.png:['super', 0.9999555]
......@@ -301,23 +297,8 @@ self.character_str = "0123456789abcdefghijklmnopqrstuvwxyz"
dict_character = list(self.character_str)
```
<a name="SRN-BASED_RECOGNITION"></a>
### 4. SRN-BASED TEXT RECOGNITION MODEL INFERENCE
The recognition model based on SRN requires additional setting of the recognition algorithm parameter --rec_algorithm="SRN".
At the same time, it is necessary to ensure that the predicted shape is consistent with the training, such as: --rec_image_shape="1, 64, 256"
```
python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words_en/word_336.png" \
--rec_model_dir="./inference/srn/" \
--rec_image_shape="1, 64, 256" \
--rec_char_type="en" \
--rec_algorithm="SRN"
```
<a name="USING_CUSTOM_CHARACTERS"></a>
### 5. TEXT RECOGNITION MODEL INFERENCE USING CUSTOM CHARACTERS DICTIONARY
### 4. TEXT RECOGNITION MODEL INFERENCE USING CUSTOM CHARACTERS DICTIONARY
If the chars dictionary is modified during training, you need to specify the new dictionary path by setting the parameter `rec_char_dict_path` when using your inference model to predict.
```
......@@ -325,7 +306,7 @@ python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words_en/word_336.png
```
<a name="MULTILINGUAL_MODEL_INFERENCE"></a>
### 6. MULTILINGAUL MODEL INFERENCE
### 5. MULTILINGAUL MODEL INFERENCE
If you need to predict other language models, when using inference model prediction, you need to specify the dictionary path used by `--rec_char_dict_path`. At the same time, in order to get the correct visualization results,
You need to specify the visual font path through `--vis_font_path`. There are small language fonts provided by default under the `doc/` path, such as Korean recognition:
......@@ -357,12 +338,14 @@ For angle classification model inference, you can execute the following commands
python3 tools/infer/predict_cls.py --image_dir="./doc/imgs_words/ch/word_4.jpg" --cls_model_dir="./inference/cls/"
```
![](../imgs_words/ch/word_4.jpg)
![](../imgs_words_en/word_10.png)
After executing the command, the prediction results (classification angle and score) of the above image will be printed on the screen.
Predicts of ./doc/imgs_words/ch/word_4.jpg:['0', 0.9999963]
```
infer_img: doc/imgs_words_en/word_10.png
result: ('0', 0.9999995)
```
<a name="CONCATENATION"></a>
## TEXT DETECTION ANGLE CLASSIFICATION AND RECOGNITION INFERENCE CONCATENATION
......
......@@ -28,21 +28,15 @@ from ppocr.modeling.architectures import build_model
from ppocr.postprocess import build_post_process
from ppocr.utils.save_load import init_model
from ppocr.utils.logging import get_logger
from tools.program import load_config
def parse_args():
parser = argparse.ArgumentParser()
parser.add_argument("-c", "--config", help="configuration file to use")
parser.add_argument(
"-o", "--output_path", type=str, default='./output/infer/')
return parser.parse_args()
from tools.program import load_config, merge_config,ArgsParser
def main():
FLAGS = parse_args()
FLAGS = ArgsParser().parse_args()
config = load_config(FLAGS.config)
merge_config(FLAGS.opt)
logger = get_logger()
print(config)
# build post process
post_process_class = build_post_process(config['PostProcess'],
......@@ -57,8 +51,7 @@ def main():
init_model(config, model, logger)
model.eval()
save_path = '{}/{}/inference'.format(FLAGS.output_path,
config['Architecture']['model_type'])
save_path = '{}/inference'.format(config['Global']['save_inference_dir'])
infer_shape = [3, 32, 100] if config['Architecture'][
'model_type'] != "det" else [3, 640, 640]
model = to_static(
......
......@@ -100,8 +100,8 @@ def create_predictor(args, mode, logger):
if model_dir is None:
logger.info("not find {} model file path {}".format(mode, model_dir))
sys.exit(0)
model_file_path = model_dir + ".pdmodel"
params_file_path = model_dir + ".pdiparams"
model_file_path = model_dir + "/inference.pdmodel"
params_file_path = model_dir + "/inference.pdiparams"
if not os.path.exists(model_file_path):
logger.info("not find model file path {}".format(model_file_path))
sys.exit(0)
......
python3 -m paddle.distributed.launch --selected_gpus '0,1,2,3,4,5,6,7' tools/train.py -c configs/rec/rec_mv3_none_bilstm_ctc.yml
\ No newline at end of file
python3 -m paddle.distributed.launch --gpus '0,1,2,3,4,5,6,7' tools/train.py -c configs/rec/rec_mv3_none_bilstm_ctc.yml
\ No newline at end of file
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册