This article provides a full-process guide for the PaddleOCR table recognition model, including data preparation, model training, tuning, evaluation, prediction, and detailed descriptions of each stage:
The default storage path for training data is `PaddleOCR/train_data`, if you already have a dataset on disk, just create a soft link to the dataset directory:
|Simple Color Table||
# 2. Training
PaddleOCR provides training scripts, evaluation scripts, and prediction scripts. In this section, the [SLANet](../../configs/table/SLANet.yml) model will be used as an example:
## 2.1. Start Training
*If you are installing the cpu version, please modify the `use_gpu` field in the configuration file to false*
```
# GPU training Support single card and multi-card training
# The training log will be automatically saved as train.log under "{save_model_dir}"
# specify the single card training(Long training time, not recommended)
The following information is automatically printed in the log:
| Field | Meaning |
| :----: | :------: |
| epoch | current iteration round |
| global_step | current iteration count |
| lr | current learning rate |
| acc | The accuracy of the current batch |
| loss | current loss function |
| structure_loss | Table Structure Loss Values |
| loc_loss | Cell Coordinate Loss Value |
| avg_reader_cost | Current batch data processing time |
| avg_batch_cost | The total time spent in the current batch |
| avg_samples | The number of samples in the current batch |
| ips | Number of images processed per second |
PaddleOCR supports alternating training and evaluation. You can modify `eval_batch_step` in `configs/table/SLANet.yml` to set the evaluation frequency. By default, it is evaluated once every 1000 iters. During the evaluation process, the best acc model is saved as `output/SLANet/best_accuracy` by default.
If the validation set is large, the test will be time-consuming. It is recommended to reduce the number of evaluations, or perform evaluation after training.
**Tips:** You can use the -c parameter to select various model configurations under the `configs/table/` path for training. For the table recognition algorithms supported by PaddleOCR, please refer to [Table Algorithms List](https://github.com/PaddlePaddle/PaddleOCR/blob/dygraph/doc/doc_en/algorithm_overview_en.md#3):
**Note that the configuration file for prediction/evaluation must be the same as training. **
## 2.2. Resume Training
If the training program is interrupted, if you want to load the interrupted model to resume training, you can specify the path of the model to be loaded by specifying Global.checkpoints:
**Note**: The priority of `Global.checkpoints` is higher than that of `Global.pretrained_model`, that is, when two parameters are specified at the same time, the model specified by `Global.checkpoints` will be loaded first. If `Global.checkpoints` The specified model path is incorrect, and the model specified by `Global.pretrained_model` will be loaded.
## 2.3. Training with New Backbone
The network part completes the construction of the network, and PaddleOCR divides the network into four parts, which are under [ppocr/modeling](../../ppocr/modeling). The data entering the network will pass through these four parts in sequence(transforms->backbones->
necks->heads).
```bash
├── architectures # Code for building network
├── transforms # Image Transformation Module
├── backbones # Feature extraction module
├── necks # Feature enhancement module
└── heads # Output module
```
If the Backbone to be replaced has a corresponding implementation in PaddleOCR, you can directly modify the parameters in the `Backbone` part of the configuration yml file.
However, if you want to use a new Backbone, an example of replacing the backbones is as follows:
1. Create a new file under the [ppocr/modeling/backbones](../../ppocr/modeling/backbones) folder, such as my_backbone.py.
2. Add code in the my_backbone.py file, the sample code is as follows:
```python
importpaddle
importpaddle.nnasnn
importpaddle.nn.functionalasF
classMyBackbone(nn.Layer):
def__init__(self,*args,**kwargs):
super(MyBackbone,self).__init__()
# your init code
self.conv=nn.xxxx
defforward(self,inputs):
# your network forward
y=self.conv(inputs)
returny
```
3. Import the added module in the [ppocr/modeling/backbones/\__init\__.py](../../ppocr/modeling/backbones/__init__.py) file.
After adding the four-part modules of the network, you only need to configure them in the configuration file to use, such as:
```yaml
Backbone:
name:MyBackbone
args1:args1
```
**NOTE**: More details about replace Backbone and other mudule can be found in [doc](add_new_algorithm_en.md).
## 2.4. Mixed Precision Training
If you want to speed up your training further, you can use [Auto Mixed Precision Training](https://www.paddlepaddle.org.cn/documentation/docs/zh/guides/01_paddle2.0_introduction/basic_concept/amp_cn.html), taking a single machine and a single gpu as an example, the commands are as follows:
During multi-machine multi-gpu training, use the `--ips` parameter to set the used machine IP address, and the `--gpus` parameter to set the used GPU ID:
**Note:** (1) When using multi-machine and multi-gpu training, you need to replace the ips value in the above command with the address of your machine, and the machines need to be able to ping each other. (2) Training needs to be launched separately on multiple machines. The command to view the ip address of the machine is `ifconfig`. (3) For more details about the distributed training speedup ratio, please refer to [Distributed Training Tutorial](./distributed_training_en.md).
## 2.6. Training with Knowledge Distillation
coming soon!
## 2.7. Training on other platform(Windows/macOS/Linux DCU)
- Windows GPU/CPU
The Windows platform is slightly different from the Linux platform:
Windows platform only supports `single gpu` training and inference, specify GPU for training `set CUDA_VISIBLE_DEVICES=0`
On the Windows platform, DataLoader only supports single-process mode, so you need to set `num_workers` to 0;
- macOS
GPU mode is not supported, you need to set `use_gpu` to False in the configuration file, and the rest of the training evaluation prediction commands are exactly the same as Linux GPU.
- Linux DCU
Running on a DCU device requires setting the environment variable `export HIP_VISIBLE_DEVICES=0,1,2,3`, and the rest of the training and evaluation prediction commands are exactly the same as the Linux GPU.
## 2.8 Fine-tuning
In the actual use process, it is recommended to load the officially provided pre-training model and fine-tune it in your own data set. For the fine-tuning method of the table recognition model, please refer to: [Model fine-tuning tutorial](./finetune.md).
# 3. Evaluation and Test
## 3.1. Evaluation
The model parameters during training are saved in the `Global.save_model_dir` directory by default. When evaluating metrics, you need to set `Global.checkpoints` to point to the saved parameter file. Evaluation datasets can be modified via the `label_file_list` setting in Eval via `configs/table/SLANet.yml`.
```
# GPU evaluation, Global.checkpoints is the weight to be tested
inference model (model saved by `paddle.jit.save`)
Generally, it is model training, a solidified model that saves the model structure and model parameters in a file, and is mostly used to predict deployment scenarios.
The model saved during the training process is the checkpoints model, and only the parameters of the model are saved, which are 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 deployment and accelerating reasoning, and is flexible and convenient, and is suitable for actual system integration.
The way to convert the form recognition model to the inference model is the same as the text detection and recognition, 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.
After the conversion is successful, there are three files in the model save directory:
```
inference/SLANet/
├── inference.pdiparams # The parameter file of inference model
├── inference.pdiparams.info # The parameter information of inference model, which can be ignored
└── inference.pdmodel # The program file of model
```
## 5. FAQ
Q1: After the training model is transferred to the inference model, the prediction effect is inconsistent?
**A**: There are many such problems, and the problems are mostly caused by inconsistent preprocessing and postprocessing parameters when the trained model predicts and the preprocessing and postprocessing parameters when the inference model predicts. You can compare whether there are differences in preprocessing, postprocessing, and prediction in the configuration files used for training.
After the operation is completed, the excel table of each image will be saved to the directory specified by the output field, and an html file will be produced in the directory to visually view the cell coordinates and the recognized table.
After the operation is completed, the excel table of each image will be saved to the directory specified by the output field, and an html file will be produced in the directory to visually view the cell coordinates and the recognized table.
- table structure recognition
```python
cdPaddleOCR/ppstructure
# download model
mkdirinference&&cdinference
# Download the PP-Structurev2 form recognition model and unzip it
After the run is complete, the visualization of the detection frame of the cell will be saved to the directory specified by the output field.
### 3.2 Train
### 3.2 Train
In this chapter, we only introduce the training of the table structure model, For model training of [text detection](../../doc/doc_en/detection_en.md) and [text recognition](../../doc/doc_en/recognition_en.md), please refer to the corresponding documents
The training, evaluation and inference process of the text detection model can be referred to [detection](../../doc/doc_en/detection_en.md)
* data preparation
For the Chinese model and the English model, the data sources are different, as follows:
English dataset: The training data uses public data set [PubTabNet](https://arxiv.org/abs/1911.10683), Can be downloaded from the official [website](https://github.com/ibm-aur-nlp/PubTabNet) 。The PubTabNet data set contains about 500,000 images, as well as annotations in html format。
Chinese dataset: The Chinese dataset consists of the following two parts, which are trained with a 1:1 sampling ratio.
> 1. Generate dataset: Use [Table Generation Tool](https://github.com/WenmuZhou/TableGeneration) to generate 40,000 images.
> 2. Crop 10,000 images from [WTW](https://github.com/wangwen-whu/WTW-Dataset).
For a detailed introduction to public datasets, please refer to [table_datasets](../../doc/doc_en/dataset/table_datasets_en.md). The following training and evaluation procedures are based on the English dataset as an example.
* Start training
*If you are installing the cpu version of paddle, please modify the `use_gpu` field in the configuration file to false*
In the above instruction, use `-c` to select the training to use the `configs/table/table_mv3.yml` configuration file.
For a detailed explanation of the configuration file, please refer to [config](../../doc/doc_en/config_en.md).
* load trained model and continue training
The training, evaluation and inference process of the text recognition model can be referred to [recognition](../../doc/doc_en/recognition_en.md)
If you expect to load trained model and continue the training again, you can specify the parameter `Global.checkpoints` as the model path to be loaded.
The training, evaluation and inference process of the table recognition model can be referred to [table_recognition](../../doc/doc_en/table_recognition_en.md)
**Note**: The priority of `Global.checkpoints` is higher than that of `Global.pretrain_weights`, that is, when two parameters are specified at the same time, the model specified by `Global.checkpoints` will be loaded first. If the model path specified by `Global.checkpoints` is wrong, the one specified by `Global.pretrain_weights` will be loaded.
### 3.3 Eval
The table uses [TEDS(Tree-Edit-Distance-based Similarity)](https://github.com/ibm-aur-nlp/PubTabNet/tree/master/src) as the evaluation metric of the model. Before the model evaluation, the three models in the pipeline need to be exported as inference models (we have provided them), and the gt for evaluation needs to be prepared. Examples of gt are as follows:
The table uses [TEDS(Tree-Edit-Distance-based Similarity)](https://github.com/ibm-aur-nlp/PubTabNet/tree/master/src) as the evaluation metric of the model. Before the model evaluation, the three models in the pipeline need to be exported as inference models (we have provided them), and the gt for evaluation needs to be prepared. Examples of gt are as follows:
