Merge pull request #4848 from WenmuZhou/vqa

add re

ppstructure/vqa/README.md

-# 视觉问答（VQA）
+# 文档视觉问答（DOC-VQA）
 
-VQA主要特性如下：
+VQA指视觉问答，主要针对图像内容进行提问和回答,DOC-VQA是VQA任务中的一种，DOC-VQA主要针对文本图像的文字内容提出问题。
+
+PP-Structure 里的 DOC-VQA算法基于PaddleNLP自然语言处理算法库进行开发。
+
+主要特性如下：
 
 - 集成[LayoutXLM](https://arxiv.org/pdf/2104.08836.pdf)模型以及PP-OCR预测引擎。
-- 支持基于多模态方法的语义实体识别 (Semantic Entity Recognition, SER) 以及关系抽取 (Relation Extraction, RE) 任务。基于 SER 任务，可以完成对图像中的文本识别与分类；基于 RE 任务，可以完成对图象中的文本内容的关系提取（比如判断问题对）
-- 支持SER任务与OCR引擎联合的端到端系统预测与评估。
-- 支持SER任务和RE任务的自定义训练
+- 支持基于多模态方法的语义实体识别 (Semantic Entity Recognition, SER) 以及关系抽取 (Relation Extraction, RE) 任务。基于 SER 任务，可以完成对图像中的文本识别与分类；基于 RE 任务，可以完成对图象中的文本内容的关系提取，如判断问题对(pair)。
+- 支持SER任务和RE任务的自定义训练。
+- 支持OCR+SER的端到端系统预测与评估。
+- 支持OCR+SER+RE的端到端系统预测。
 
 
 本项目是 [LayoutXLM: Multimodal Pre-training for Multilingual Visually-rich Document Understanding](https://arxiv.org/pdf/2104.08836.pdf) 在 Paddle 2.2上的开源实现，
 包含了在 [XFUND数据集](https://github.com/doc-analysis/XFUND) 上的微调代码。
 
-## 1. 效果演示
+## 1 性能
+
+我们在 [XFUN](https://github.com/doc-analysis/XFUND) 评估数据集上对算法进行了评估，性能如下
+
+|任务|    f1 | 模型下载地址|
+|:---:|:---:| :---:|
+|SER|0.9056| [链接](https://paddleocr.bj.bcebos.com/pplayout/PP-Layout_v1.0_ser_pretrained.tar)|
+|RE|0.7113| [链接](https://paddleocr.bj.bcebos.com/pplayout/PP-Layout_v1.0_re_pretrained.tar)|
+
+
+
+## 2. 效果演示
 
 **注意：** 测试图片来源于XFUN数据集。
 
-### 1.1 SER
+### 2.1 SER
 
-<div align="center">
-<img src="./images/result_ser/zh_val_0_ser.jpg"  width = "600" />
-</div>
+![](./images/result_ser/zh_val_0_ser.jpg) | ![](./images/result_ser/zh_val_42_ser.jpg)
+---|---
 
-<div align="center">
-<img src="./images/result_ser/zh_val_42_ser.jpg"  width = "600" />
-</div>
+图中不同颜色的框表示不同的类别，对于XFUN数据集，有`QUESTION`, `ANSWER`, `HEADER` 3种类别
 
-其中不同颜色的框表示不同的类别，对于XFUN数据集，有`QUESTION`, `ANSWER`, `HEADER` 3种类别，在OCR检测框的左上方也标出了对应的类别和OCR识别结果。
+* 深紫色：HEADER
+* 浅紫色：QUESTION
+* 军绿色：ANSWER
 
+在OCR检测框的左上方也标出了对应的类别和OCR识别结果。
 
-### 1.2 RE
 
-* Coming soon!
+### 2.2 RE
 
+![](./images/result_re/zh_val_21_re.jpg) | ![](./images/result_re/zh_val_40_re.jpg)
+---|---
 
 
-## 2. 安装
+图中红色框表示问题，蓝色框表示答案，问题和答案之间使用绿色线连接。在OCR检测框的左上方也标出了对应的类别和OCR识别结果。
 
-### 2.1 安装依赖
+
+## 3. 安装
+
+### 3.1 安装依赖
 
 - **（1) 安装PaddlePaddle**
 
@@ -53,12 +73,12 @@ python3 -m pip install paddlepaddle==2.2 -i https://mirror.baidu.com/pypi/simple
 更多需求，请参照[安装文档](https://www.paddlepaddle.org.cn/install/quick)中的说明进行操作。
 
 
-### 2.2 安装PaddleOCR（包含 PP-OCR 和 VQA ）
+### 3.2 安装PaddleOCR（包含 PP-OCR 和 VQA ）
 
 - **（1）pip快速安装PaddleOCR whl包（仅预测）**
 
 ```bash
-pip install "paddleocr>=2.2" # 推荐使用2.2+版本
+pip install paddleocr
 ```
 
 - **（2）下载VQA源码（预测+训练）**
@@ -85,13 +105,14 @@ pip install -e .
 - **（4）安装VQA的`requirements`**
 
 ```bash
+cd ppstructure/vqa
 pip install -r requirements.txt
 ```
 
-## 3. 使用
+## 4. 使用
 
 
-### 3.1 数据和预训练模型准备
+### 4.1 数据和预训练模型准备
 
 处理好的XFUN中文数据集下载地址：[https://paddleocr.bj.bcebos.com/dataset/XFUND.tar](https://paddleocr.bj.bcebos.com/dataset/XFUND.tar)。
 
@@ -104,18 +125,15 @@ wget https://paddleocr.bj.bcebos.com/dataset/XFUND.tar
 
 如果希望转换XFUN中其他语言的数据集，可以参考[XFUN数据转换脚本](helper/trans_xfun_data.py)。
 
-如果希望直接体验预测过程，可以下载我们提供的SER预训练模型，跳过训练过程，直接预测即可。
-
-* SER任务预训练模型下载链接：[链接](https://paddleocr.bj.bcebos.com/pplayout/PP-Layout_v1.0_ser_pretrained.tar)
-* RE任务预训练模型下载链接：coming soon!
+如果希望直接体验预测过程，可以下载我们提供的预训练模型，跳过训练过程，直接预测即可。
 
 
-### 3.2 SER任务
+### 4.2 SER任务
 
 * 启动训练
 
 ```shell
-python train_ser.py \
+python3.7 train_ser.py \
     --model_name_or_path "layoutxlm-base-uncased" \
     --train_data_dir "XFUND/zh_train/image" \
     --train_label_path "XFUND/zh_train/xfun_normalize_train.json" \
@@ -131,13 +149,7 @@ python train_ser.py \
     --seed 2048
 ```
 
-最终会打印出`precision`, `recall`, `f1`等指标，如下所示。
-
-```
-best metrics: {'loss': 1.066644651549203, 'precision': 0.8770182068017863, 'recall': 0.9361936193619362, 'f1': 0.9056402979780063}
-```
-
-模型和训练日志会保存在`./output/ser/`文件夹中。
+最终会打印出`precision`, `recall`, `f1`等指标，模型和训练日志会保存在`./output/ser/`文件夹中。
 
 * 使用评估集合中提供的OCR识别结果进行预测
 
@@ -159,21 +171,73 @@ export CUDA_VISIBLE_DEVICES=0
 python3.7 infer_ser_e2e.py \
     --model_name_or_path "./output/PP-Layout_v1.0_ser_pretrained/" \
     --max_seq_length 512 \
-    --output_dir "output_res_e2e/"
+    --output_dir "output_res_e2e/" \
+    --infer_imgs "images/input/zh_val_0.jpg"
 ```
 
 * 对`OCR引擎 + SER`预测系统进行端到端评估
 
 ```shell
 export CUDA_VISIBLE_DEVICES=0
-python helper/eval_with_label_end2end.py --gt_json_path XFUND/zh_val/xfun_normalize_val.json  --pred_json_path output_res/infer_results.txt
+python3.7 helper/eval_with_label_end2end.py --gt_json_path XFUND/zh_val/xfun_normalize_val.json  --pred_json_path output_res/infer_results.txt
 ```
 
 
-3.3 RE任务
+### 3.3 RE任务
 
-coming soon!
+* 启动训练
 
+```shell
+python3 train_re.py \
+    --model_name_or_path "layoutxlm-base-uncased" \
+    --train_data_dir "XFUND/zh_train/image" \
+    --train_label_path "XFUND/zh_train/xfun_normalize_train.json" \
+    --eval_data_dir "XFUND/zh_val/image" \
+    --eval_label_path "XFUND/zh_val/xfun_normalize_val.json" \
+    --label_map_path 'labels/labels_ser.txt' \
+    --num_train_epochs 2 \
+    --eval_steps 10 \
+    --save_steps 500 \
+    --output_dir "output/re/"  \
+    --learning_rate 5e-5 \
+    --warmup_steps 50 \
+    --per_gpu_train_batch_size 8 \
+    --per_gpu_eval_batch_size 8 \
+    --evaluate_during_training \
+    --seed 2048
+
+```
+
+最终会打印出`precision`, `recall`, `f1`等指标，模型和训练日志会保存在`./output/re/`文件夹中。
+
+* 使用评估集合中提供的OCR识别结果进行预测
+
+```shell
+export CUDA_VISIBLE_DEVICES=0
+python3 infer_re.py \
+    --model_name_or_path "./PP-Layout_v1.0_re_pretrained/" \
+    --max_seq_length 512 \
+    --eval_data_dir "XFUND/zh_val/image" \
+    --eval_label_path "XFUND/zh_val/xfun_normalize_val.json" \
+    --label_map_path 'labels/labels_ser.txt' \
+    --output_dir "output_res"  \
+    --per_gpu_eval_batch_size 1 \
+    --seed 2048
+```
+
+最终会在`output_res`目录下保存预测结果可视化图像以及预测结果文本文件，文件名为`infer_results.txt`。
+
+* 使用`OCR引擎 + SER + RE`串联结果
+
+```shell
+export CUDA_VISIBLE_DEVICES=0
+# python3.7 infer_ser_re_e2e.py \
+    --model_name_or_path "./PP-Layout_v1.0_ser_pretrained/" \
+    --re_model_name_or_path "./PP-Layout_v1.0_re_pretrained/" \
+    --max_seq_length 512 \
+    --output_dir "output_ser_re_e2e_train/" \
+    --infer_imgs "images/input/zh_val_21.jpg"
+```
 
 ## 参考链接
 

ppstructure/vqa/data_collator.py

+# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import paddle
+import numbers
+import numpy as np
+
+
+class DataCollator:
+    """
+    data batch
+    """
+
+    def __call__(self, batch):
+        data_dict = {}
+        to_tensor_keys = []
+        for sample in batch:
+            for k, v in sample.items():
+                if k not in data_dict:
+                    data_dict[k] = []
+                if isinstance(v, (np.ndarray, paddle.Tensor, numbers.Number)):
+                    if k not in to_tensor_keys:
+                        to_tensor_keys.append(k)
+                data_dict[k].append(v)
+        for k in to_tensor_keys:
+            data_dict[k] = paddle.to_tensor(data_dict[k])
+        return data_dict

ppstructure/vqa/infer_re.py

+import os
+import sys
+
+__dir__ = os.path.dirname(os.path.abspath(__file__))
+sys.path.append(__dir__)
+sys.path.append(os.path.abspath(os.path.join(__dir__, '../..')))
+
+import random
+
+import cv2
+import matplotlib.pyplot as plt
+import numpy as np
+import paddle
+
+from paddlenlp.transformers import LayoutXLMTokenizer, LayoutXLMModel, LayoutXLMForRelationExtraction
+
+from xfun import XFUNDataset
+from utils import parse_args, get_bio_label_maps, draw_re_results
+from data_collator import DataCollator
+
+from ppocr.utils.logging import get_logger
+
+
+def infer(args):
+    os.makedirs(args.output_dir, exist_ok=True)
+    logger = get_logger()
+    label2id_map, id2label_map = get_bio_label_maps(args.label_map_path)
+    pad_token_label_id = paddle.nn.CrossEntropyLoss().ignore_index
+
+    tokenizer = LayoutXLMTokenizer.from_pretrained(args.model_name_or_path)
+
+    model = LayoutXLMForRelationExtraction.from_pretrained(
+        args.model_name_or_path)
+
+    eval_dataset = XFUNDataset(
+        tokenizer,
+        data_dir=args.eval_data_dir,
+        label_path=args.eval_label_path,
+        label2id_map=label2id_map,
+        img_size=(224, 224),
+        max_seq_len=args.max_seq_length,
+        pad_token_label_id=pad_token_label_id,
+        contains_re=True,
+        add_special_ids=False,
+        return_attention_mask=True,
+        load_mode='all')
+
+    eval_dataloader = paddle.io.DataLoader(
+        eval_dataset,
+        batch_size=args.per_gpu_eval_batch_size,
+        num_workers=8,
+        shuffle=False,
+        collate_fn=DataCollator())
+
+    # 读取gt的oct数据
+    ocr_info_list = load_ocr(args.eval_data_dir, args.eval_label_path)
+
+    for idx, batch in enumerate(eval_dataloader):
+        logger.info("[Infer] process: {}/{}".format(idx, len(eval_dataloader)))
+        with paddle.no_grad():
+            outputs = model(**batch)
+        pred_relations = outputs['pred_relations']
+
+        ocr_info = ocr_info_list[idx]
+        image_path = ocr_info['image_path']
+        ocr_info = ocr_info['ocr_info']
+
+        # 根据entity里的信息，做token解码后去过滤不要的ocr_info
+        ocr_info = filter_bg_by_txt(ocr_info, batch, tokenizer)
+
+        # 进行 relations 到 ocr信息的转换
+        result = []
+        used_tail_id = []
+        for relations in pred_relations:
+            for relation in relations:
+                if relation['tail_id'] in used_tail_id:
+                    continue
+                if relation['head_id'] not in ocr_info or relation[
+                        'tail_id'] not in ocr_info:
+                    continue
+                used_tail_id.append(relation['tail_id'])
+                ocr_info_head = ocr_info[relation['head_id']]
+                ocr_info_tail = ocr_info[relation['tail_id']]
+                result.append((ocr_info_head, ocr_info_tail))
+
+        img = cv2.imread(image_path)
+        img_show = draw_re_results(img, result)
+        save_path = os.path.join(args.output_dir, os.path.basename(image_path))
+        cv2.imwrite(save_path, img_show)
+
+
+def load_ocr(img_folder, json_path):
+    import json
+    d = []
+    with open(json_path, "r") as fin:
+        lines = fin.readlines()
+        for line in lines:
+            image_name, info_str = line.split("\t")
+            info_dict = json.loads(info_str)
+            info_dict['image_path'] = os.path.join(img_folder, image_name)
+            d.append(info_dict)
+    return d
+
+
+def filter_bg_by_txt(ocr_info, batch, tokenizer):
+    entities = batch['entities'][0]
+    input_ids = batch['input_ids'][0]
+
+    new_info_dict = {}
+    for i in range(len(entities['start'])):
+        entitie_head = entities['start'][i]
+        entitie_tail = entities['end'][i]
+        word_input_ids = input_ids[entitie_head:entitie_tail].numpy().tolist()
+        txt = tokenizer.convert_ids_to_tokens(word_input_ids)
+        txt = tokenizer.convert_tokens_to_string(txt)
+
+        for i, info in enumerate(ocr_info):
+            if info['text'] == txt:
+                new_info_dict[i] = info
+    return new_info_dict
+
+
+def post_process(pred_relations, ocr_info, img):
+    result = []
+    for relations in pred_relations:
+        for relation in relations:
+            ocr_info_head = ocr_info[relation['head_id']]
+            ocr_info_tail = ocr_info[relation['tail_id']]
+            result.append((ocr_info_head, ocr_info_tail))
+    return result
+
+
+def draw_re(result, image_path, output_folder):
+    img = cv2.imread(image_path)
+
+    from matplotlib import pyplot as plt
+    for ocr_info_head, ocr_info_tail in result:
+        cv2.rectangle(
+            img,
+            tuple(ocr_info_head['bbox'][:2]),
+            tuple(ocr_info_head['bbox'][2:]), (255, 0, 0),
+            thickness=2)
+        cv2.rectangle(
+            img,
+            tuple(ocr_info_tail['bbox'][:2]),
+            tuple(ocr_info_tail['bbox'][2:]), (0, 0, 255),
+            thickness=2)
+        center_p1 = [(ocr_info_head['bbox'][0] + ocr_info_head['bbox'][2]) // 2,
+                     (ocr_info_head['bbox'][1] + ocr_info_head['bbox'][3]) // 2]
+        center_p2 = [(ocr_info_tail['bbox'][0] + ocr_info_tail['bbox'][2]) // 2,
+                     (ocr_info_tail['bbox'][1] + ocr_info_tail['bbox'][3]) // 2]
+        cv2.line(
+            img, tuple(center_p1), tuple(center_p2), (0, 255, 0), thickness=2)
+    plt.imshow(img)
+    plt.savefig(
+        os.path.join(output_folder, os.path.basename(image_path)), dpi=600)
+    # plt.show()
+
+
+if __name__ == "__main__":
+    args = parse_args()
+    infer(args)

ppstructure/vqa/infer_ser_e2e.py

@@ -23,8 +23,10 @@ from PIL import Image
 import paddle
 from paddlenlp.transformers import LayoutXLMModel, LayoutXLMTokenizer, LayoutXLMForTokenClassification
 
+from paddleocr import PaddleOCR
+
 # relative reference
-from utils import parse_args, get_image_file_list, draw_ser_results, get_bio_label_maps, build_ocr_engine
+from utils import parse_args, get_image_file_list, draw_ser_results, get_bio_label_maps
 
 from utils import pad_sentences, split_page, preprocess, postprocess, merge_preds_list_with_ocr_info
 
@@ -48,74 +50,82 @@ def parse_ocr_info_for_ser(ocr_result):
     return ocr_info
 
 
-@paddle.no_grad()
-def infer(args):
-    os.makedirs(args.output_dir, exist_ok=True)
+class SerPredictor(object):
+    def __init__(self, args):
+        self.max_seq_length = args.max_seq_length
+
+        # init ser token and model
+        self.tokenizer = LayoutXLMTokenizer.from_pretrained(
+            args.model_name_or_path)
+        self.model = LayoutXLMForTokenClassification.from_pretrained(
+            args.model_name_or_path)
+        self.model.eval()
+
+        # init ocr_engine
+        self.ocr_engine = PaddleOCR(
+            rec_model_dir=args.ocr_rec_model_dir,
+            det_model_dir=args.ocr_det_model_dir,
+            use_angle_cls=False,
+            show_log=False)
+        # init dict
+        label2id_map, self.id2label_map = get_bio_label_maps(
+            args.label_map_path)
+        self.label2id_map_for_draw = dict()
+        for key in label2id_map:
+            if key.startswith("I-"):
+                self.label2id_map_for_draw[key] = label2id_map["B" + key[1:]]
+            else:
+                self.label2id_map_for_draw[key] = label2id_map[key]
+
+    def __call__(self, img):
+        ocr_result = self.ocr_engine.ocr(img, cls=False)
+
+        ocr_info = parse_ocr_info_for_ser(ocr_result)
+
+        inputs = preprocess(
+            tokenizer=self.tokenizer,
+            ori_img=img,
+            ocr_info=ocr_info,
+            max_seq_len=self.max_seq_length)
+
+        outputs = self.model(
+            input_ids=inputs["input_ids"],
+            bbox=inputs["bbox"],
+            image=inputs["image"],
+            token_type_ids=inputs["token_type_ids"],
+            attention_mask=inputs["attention_mask"])
+
+        preds = outputs[0]
+        preds = postprocess(inputs["attention_mask"], preds, self.id2label_map)
+        ocr_info = merge_preds_list_with_ocr_info(
+            ocr_info, inputs["segment_offset_id"], preds,
+            self.label2id_map_for_draw)
+        return ocr_info, inputs
 
-    # init token and model
-    tokenizer = LayoutXLMTokenizer.from_pretrained(args.model_name_or_path)
-    model = LayoutXLMForTokenClassification.from_pretrained(
-        args.model_name_or_path)
-    model.eval()
 
-    label2id_map, id2label_map = get_bio_label_maps(args.label_map_path)
-    label2id_map_for_draw = dict()
-    for key in label2id_map:
-        if key.startswith("I-"):
-            label2id_map_for_draw[key] = label2id_map["B" + key[1:]]
-        else:
-            label2id_map_for_draw[key] = label2id_map[key]
+if __name__ == "__main__":
+    args = parse_args()
+    os.makedirs(args.output_dir, exist_ok=True)
 
     # get infer img list
     infer_imgs = get_image_file_list(args.infer_imgs)
 
-    ocr_engine = build_ocr_engine(args.ocr_rec_model_dir,
-                                  args.ocr_det_model_dir)
-
     # loop for infer
+    ser_engine = SerPredictor(args)
     with open(os.path.join(args.output_dir, "infer_results.txt"), "w") as fout:
         for idx, img_path in enumerate(infer_imgs):
-            print("process: [{}/{}]".format(idx, len(infer_imgs), img_path))
+            print("process: [{}/{}], {}".format(idx, len(infer_imgs), img_path))
 
             img = cv2.imread(img_path)
 
-            ocr_result = ocr_engine.ocr(img_path, cls=False)
-
-            ocr_info = parse_ocr_info_for_ser(ocr_result)
-
-            inputs = preprocess(
-                tokenizer=tokenizer,
-                ori_img=img,
-                ocr_info=ocr_info,
-                max_seq_len=args.max_seq_length)
-
-            outputs = model(
-                input_ids=inputs["input_ids"],
-                bbox=inputs["bbox"],
-                image=inputs["image"],
-                token_type_ids=inputs["token_type_ids"],
-                attention_mask=inputs["attention_mask"])
-
-            preds = outputs[0]
-            preds = postprocess(inputs["attention_mask"], preds, id2label_map)
-            ocr_info = merge_preds_list_with_ocr_info(
-                ocr_info, inputs["segment_offset_id"], preds,
-                label2id_map_for_draw)
-
+            result, _ = ser_engine(img)
             fout.write(img_path + "\t" + json.dumps(
                 {
-                    "ocr_info": ocr_info,
+                    "ser_resule": result,
                 }, ensure_ascii=False) + "\n")
 
-            img_res = draw_ser_results(img, ocr_info)
+            img_res = draw_ser_results(img, result)
             cv2.imwrite(
                 os.path.join(args.output_dir,
                              os.path.splitext(os.path.basename(img_path))[0] +
                              "_ser.jpg"), img_res)
-
-    return
-
-
-if __name__ == "__main__":
-    args = parse_args()
-    infer(args)

ppstructure/vqa/infer_ser_re_e2e.py

+# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import os
+import sys
+import json
+import cv2
+import numpy as np
+from copy import deepcopy
+from PIL import Image
+
+import paddle
+from paddlenlp.transformers import LayoutXLMModel, LayoutXLMTokenizer, LayoutXLMForRelationExtraction
+
+# relative reference
+from utils import parse_args, get_image_file_list, draw_re_results
+from infer_ser_e2e import SerPredictor
+
+
+def make_input(ser_input, ser_result, max_seq_len=512):
+    entities_labels = {'HEADER': 0, 'QUESTION': 1, 'ANSWER': 2}
+
+    entities = ser_input['entities'][0]
+    assert len(entities) == len(ser_result)
+
+    # entities
+    start = []
+    end = []
+    label = []
+    entity_idx_dict = {}
+    for i, (res, entity) in enumerate(zip(ser_result, entities)):
+        if res['pred'] == 'O':
+            continue
+        entity_idx_dict[len(start)] = i
+        start.append(entity['start'])
+        end.append(entity['end'])
+        label.append(entities_labels[res['pred']])
+    entities = dict(start=start, end=end, label=label)
+
+    # relations
+    head = []
+    tail = []
+    for i in range(len(entities["label"])):
+        for j in range(len(entities["label"])):
+            if entities["label"][i] == 1 and entities["label"][j] == 2:
+                head.append(i)
+                tail.append(j)
+
+    relations = dict(head=head, tail=tail)
+
+    batch_size = ser_input["input_ids"].shape[0]
+    entities_batch = []
+    relations_batch = []
+    for b in range(batch_size):
+        entities_batch.append(entities)
+        relations_batch.append(relations)
+
+    ser_input['entities'] = entities_batch
+    ser_input['relations'] = relations_batch
+
+    ser_input.pop('segment_offset_id')
+    return ser_input, entity_idx_dict
+
+
+class SerReSystem(object):
+    def __init__(self, args):
+        self.ser_engine = SerPredictor(args)
+        self.tokenizer = LayoutXLMTokenizer.from_pretrained(
+            args.re_model_name_or_path)
+        self.model = LayoutXLMForRelationExtraction.from_pretrained(
+            args.re_model_name_or_path)
+        self.model.eval()
+
+    def __call__(self, img):
+        ser_result, ser_inputs = self.ser_engine(img)
+        re_input, entity_idx_dict = make_input(ser_inputs, ser_result)
+
+        re_result = self.model(**re_input)
+
+        pred_relations = re_result['pred_relations'][0]
+        # 进行 relations 到 ocr信息的转换
+        result = []
+        used_tail_id = []
+        for relation in pred_relations:
+            if relation['tail_id'] in used_tail_id:
+                continue
+            used_tail_id.append(relation['tail_id'])
+            ocr_info_head = ser_result[entity_idx_dict[relation['head_id']]]
+            ocr_info_tail = ser_result[entity_idx_dict[relation['tail_id']]]
+            result.append((ocr_info_head, ocr_info_tail))
+
+        return result
+
+
+if __name__ == "__main__":
+    args = parse_args()
+    os.makedirs(args.output_dir, exist_ok=True)
+
+    # get infer img list
+    infer_imgs = get_image_file_list(args.infer_imgs)
+
+    # loop for infer
+    ser_re_engine = SerReSystem(args)
+    with open(os.path.join(args.output_dir, "infer_results.txt"), "w") as fout:
+        for idx, img_path in enumerate(infer_imgs):
+            print("process: [{}/{}], {}".format(idx, len(infer_imgs), img_path))
+
+            img = cv2.imread(img_path)
+
+            result = ser_re_engine(img)
+            fout.write(img_path + "\t" + json.dumps(
+                {
+                    "result": result,
+                }, ensure_ascii=False) + "\n")
+
+            img_res = draw_re_results(img, result)
+            cv2.imwrite(
+                os.path.join(args.output_dir,
+                             os.path.splitext(os.path.basename(img_path))[0] +
+                             "_re.jpg"), img_res)

ppstructure/vqa/metric.py

+# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import os
+import re
+
+import numpy as np
+
+import logging
+
+logger = logging.getLogger(__name__)
+
+PREFIX_CHECKPOINT_DIR = "checkpoint"
+_re_checkpoint = re.compile(r"^" + PREFIX_CHECKPOINT_DIR + r"\-(\d+)$")
+
+
+def get_last_checkpoint(folder):
+    content = os.listdir(folder)
+    checkpoints = [
+        path for path in content
+        if _re_checkpoint.search(path) is not None and os.path.isdir(
+            os.path.join(folder, path))
+    ]
+    if len(checkpoints) == 0:
+        return
+    return os.path.join(
+        folder,
+        max(checkpoints,
+            key=lambda x: int(_re_checkpoint.search(x).groups()[0])))
+
+
+def re_score(pred_relations, gt_relations, mode="strict"):
+    """Evaluate RE predictions
+
+    Args:
+        pred_relations (list) :  list of list of predicted relations (several relations in each sentence)
+        gt_relations (list) :    list of list of ground truth relations
+
+            rel = { "head": (start_idx (inclusive), end_idx (exclusive)),
+                    "tail": (start_idx (inclusive), end_idx (exclusive)),
+                    "head_type": ent_type,
+                    "tail_type": ent_type,
+                    "type": rel_type}
+
+        vocab (Vocab) :         dataset vocabulary
+        mode (str) :            in 'strict' or 'boundaries'"""
+
+    assert mode in ["strict", "boundaries"]
+
+    relation_types = [v for v in [0, 1] if not v == 0]
+    scores = {
+        rel: {
+            "tp": 0,
+            "fp": 0,
+            "fn": 0
+        }
+        for rel in relation_types + ["ALL"]
+    }
+
+    # Count GT relations and Predicted relations
+    n_sents = len(gt_relations)
+    n_rels = sum([len([rel for rel in sent]) for sent in gt_relations])
+    n_found = sum([len([rel for rel in sent]) for sent in pred_relations])
+
+    # Count TP, FP and FN per type
+    for pred_sent, gt_sent in zip(pred_relations, gt_relations):
+        for rel_type in relation_types:
+            # strict mode takes argument types into account
+            if mode == "strict":
+                pred_rels = {(rel["head"], rel["head_type"], rel["tail"],
+                              rel["tail_type"])
+                             for rel in pred_sent if rel["type"] == rel_type}
+                gt_rels = {(rel["head"], rel["head_type"], rel["tail"],
+                            rel["tail_type"])
+                           for rel in gt_sent if rel["type"] == rel_type}
+
+            # boundaries mode only takes argument spans into account
+            elif mode == "boundaries":
+                pred_rels = {(rel["head"], rel["tail"])
+                             for rel in pred_sent if rel["type"] == rel_type}
+                gt_rels = {(rel["head"], rel["tail"])
+                           for rel in gt_sent if rel["type"] == rel_type}
+
+            scores[rel_type]["tp"] += len(pred_rels & gt_rels)
+            scores[rel_type]["fp"] += len(pred_rels - gt_rels)
+            scores[rel_type]["fn"] += len(gt_rels - pred_rels)
+
+    # Compute per entity Precision / Recall / F1
+    for rel_type in scores.keys():
+        if scores[rel_type]["tp"]:
+            scores[rel_type]["p"] = scores[rel_type]["tp"] / (
+                scores[rel_type]["fp"] + scores[rel_type]["tp"])
+            scores[rel_type]["r"] = scores[rel_type]["tp"] / (
+                scores[rel_type]["fn"] + scores[rel_type]["tp"])
+        else:
+            scores[rel_type]["p"], scores[rel_type]["r"] = 0, 0
+
+        if not scores[rel_type]["p"] + scores[rel_type]["r"] == 0:
+            scores[rel_type]["f1"] = (
+                2 * scores[rel_type]["p"] * scores[rel_type]["r"] /
+                (scores[rel_type]["p"] + scores[rel_type]["r"]))
+        else:
+            scores[rel_type]["f1"] = 0
+
+    # Compute micro F1 Scores
+    tp = sum([scores[rel_type]["tp"] for rel_type in relation_types])
+    fp = sum([scores[rel_type]["fp"] for rel_type in relation_types])
+    fn = sum([scores[rel_type]["fn"] for rel_type in relation_types])
+
+    if tp:
+        precision = tp / (tp + fp)
+        recall = tp / (tp + fn)
+        f1 = 2 * precision * recall / (precision + recall)
+
+    else:
+        precision, recall, f1 = 0, 0, 0
+
+    scores["ALL"]["p"] = precision
+    scores["ALL"]["r"] = recall
+    scores["ALL"]["f1"] = f1
+    scores["ALL"]["tp"] = tp
+    scores["ALL"]["fp"] = fp
+    scores["ALL"]["fn"] = fn
+
+    # Compute Macro F1 Scores
+    scores["ALL"]["Macro_f1"] = np.mean(
+        [scores[ent_type]["f1"] for ent_type in relation_types])
+    scores["ALL"]["Macro_p"] = np.mean(
+        [scores[ent_type]["p"] for ent_type in relation_types])
+    scores["ALL"]["Macro_r"] = np.mean(
+        [scores[ent_type]["r"] for ent_type in relation_types])
+
+    # logger.info(f"RE Evaluation in *** {mode.upper()} *** mode")
+
+    # logger.info(
+    #     "processed {} sentences with {} relations; found: {} relations; correct: {}.".format(
+    #         n_sents, n_rels, n_found, tp
+    #     )
+    # )
+    # logger.info(
+    #     "\tALL\t TP: {};\tFP: {};\tFN: {}".format(scores["ALL"]["tp"], scores["ALL"]["fp"], scores["ALL"]["fn"])
+    # )
+    # logger.info("\t\t(m avg): precision: {:.2f};\trecall: {:.2f};\tf1: {:.2f} (micro)".format(precision, recall, f1))
+    # logger.info(
+    #     "\t\t(M avg): precision: {:.2f};\trecall: {:.2f};\tf1: {:.2f} (Macro)\n".format(
+    #         scores["ALL"]["Macro_p"], scores["ALL"]["Macro_r"], scores["ALL"]["Macro_f1"]
+    #     )
+    # )
+
+    # for rel_type in relation_types:
+    #     logger.info(
+    #         "\t{}: \tTP: {};\tFP: {};\tFN: {};\tprecision: {:.2f};\trecall: {:.2f};\tf1: {:.2f};\t{}".format(
+    #             rel_type,
+    #             scores[rel_type]["tp"],
+    #             scores[rel_type]["fp"],
+    #             scores[rel_type]["fn"],
+    #             scores[rel_type]["p"],
+    #             scores[rel_type]["r"],
+    #             scores[rel_type]["f1"],
+    #             scores[rel_type]["tp"] + scores[rel_type]["fp"],
+    #         )
+    #     )
+
+    return scores

ppstructure/vqa/train_re.py

+# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import os
+import sys
+
+__dir__ = os.path.dirname(os.path.abspath(__file__))
+sys.path.append(__dir__)
+sys.path.append(os.path.abspath(os.path.join(__dir__, '../..')))
+
+import random
+import numpy as np
+import paddle
+
+from paddlenlp.transformers import LayoutXLMTokenizer, LayoutXLMModel, LayoutXLMForRelationExtraction
+
+from xfun import XFUNDataset
+from utils import parse_args, get_bio_label_maps, print_arguments
+from data_collator import DataCollator
+from metric import re_score
+
+from ppocr.utils.logging import get_logger
+
+
+def set_seed(seed):
+    random.seed(seed)
+    np.random.seed(seed)
+    paddle.seed(seed)
+
+
+def cal_metric(re_preds, re_labels, entities):
+    gt_relations = []
+    for b in range(len(re_labels)):
+        rel_sent = []
+        for head, tail in zip(re_labels[b]["head"], re_labels[b]["tail"]):
+            rel = {}
+            rel["head_id"] = head
+            rel["head"] = (entities[b]["start"][rel["head_id"]],
+                           entities[b]["end"][rel["head_id"]])
+            rel["head_type"] = entities[b]["label"][rel["head_id"]]
+
+            rel["tail_id"] = tail
+            rel["tail"] = (entities[b]["start"][rel["tail_id"]],
+                           entities[b]["end"][rel["tail_id"]])
+            rel["tail_type"] = entities[b]["label"][rel["tail_id"]]
+
+            rel["type"] = 1
+            rel_sent.append(rel)
+        gt_relations.append(rel_sent)
+    re_metrics = re_score(re_preds, gt_relations, mode="boundaries")
+    return re_metrics
+
+
+def evaluate(model, eval_dataloader, logger, prefix=""):
+    # Eval!
+    logger.info("***** Running evaluation {} *****".format(prefix))
+    logger.info("  Num examples = {}".format(len(eval_dataloader.dataset)))
+
+    re_preds = []
+    re_labels = []
+    entities = []
+    eval_loss = 0.0
+    model.eval()
+    for idx, batch in enumerate(eval_dataloader):
+        with paddle.no_grad():
+            outputs = model(**batch)
+            loss = outputs['loss'].mean().item()
+            if paddle.distributed.get_rank() == 0:
+                logger.info("[Eval] process: {}/{}, loss: {:.5f}".format(
+                    idx, len(eval_dataloader), loss))
+
+            eval_loss += loss
+        re_preds.extend(outputs['pred_relations'])
+        re_labels.extend(batch['relations'])
+        entities.extend(batch['entities'])
+    re_metrics = cal_metric(re_preds, re_labels, entities)
+    re_metrics = {
+        "precision": re_metrics["ALL"]["p"],
+        "recall": re_metrics["ALL"]["r"],
+        "f1": re_metrics["ALL"]["f1"],
+    }
+    model.train()
+    return re_metrics
+
+
+def train(args):
+    logger = get_logger(log_file=os.path.join(args.output_dir, "train.log"))
+    print_arguments(args, logger)
+
+    # Added here for reproducibility (even between python 2 and 3)
+    set_seed(args.seed)
+
+    label2id_map, id2label_map = get_bio_label_maps(args.label_map_path)
+    pad_token_label_id = paddle.nn.CrossEntropyLoss().ignore_index
+
+    # dist mode
+    if paddle.distributed.get_world_size() > 1:
+        paddle.distributed.init_parallel_env()
+
+    tokenizer = LayoutXLMTokenizer.from_pretrained(args.model_name_or_path)
+
+    model = LayoutXLMModel.from_pretrained(args.model_name_or_path)
+    model = LayoutXLMForRelationExtraction(model, dropout=None)
+
+    # dist mode
+    if paddle.distributed.get_world_size() > 1:
+        model = paddle.distributed.DataParallel(model)
+
+    train_dataset = XFUNDataset(
+        tokenizer,
+        data_dir=args.train_data_dir,
+        label_path=args.train_label_path,
+        label2id_map=label2id_map,
+        img_size=(224, 224),
+        max_seq_len=args.max_seq_length,
+        pad_token_label_id=pad_token_label_id,
+        contains_re=True,
+        add_special_ids=False,
+        return_attention_mask=True,
+        load_mode='all')
+
+    eval_dataset = XFUNDataset(
+        tokenizer,
+        data_dir=args.eval_data_dir,
+        label_path=args.eval_label_path,
+        label2id_map=label2id_map,
+        img_size=(224, 224),
+        max_seq_len=args.max_seq_length,
+        pad_token_label_id=pad_token_label_id,
+        contains_re=True,
+        add_special_ids=False,
+        return_attention_mask=True,
+        load_mode='all')
+
+    train_sampler = paddle.io.DistributedBatchSampler(
+        train_dataset, batch_size=args.per_gpu_train_batch_size, shuffle=True)
+    args.train_batch_size = args.per_gpu_train_batch_size * \
+                            max(1, paddle.distributed.get_world_size())
+    train_dataloader = paddle.io.DataLoader(
+        train_dataset,
+        batch_sampler=train_sampler,
+        num_workers=8,
+        use_shared_memory=True,
+        collate_fn=DataCollator())
+
+    eval_dataloader = paddle.io.DataLoader(
+        eval_dataset,
+        batch_size=args.per_gpu_eval_batch_size,
+        num_workers=8,
+        shuffle=False,
+        collate_fn=DataCollator())
+
+    t_total = len(train_dataloader) * args.num_train_epochs
+
+    # build linear decay with warmup lr sch
+    lr_scheduler = paddle.optimizer.lr.PolynomialDecay(
+        learning_rate=args.learning_rate,
+        decay_steps=t_total,
+        end_lr=0.0,
+        power=1.0)
+    if args.warmup_steps > 0:
+        lr_scheduler = paddle.optimizer.lr.LinearWarmup(
+            lr_scheduler,
+            args.warmup_steps,
+            start_lr=0,
+            end_lr=args.learning_rate, )
+    grad_clip = paddle.nn.ClipGradByNorm(clip_norm=10)
+    optimizer = paddle.optimizer.Adam(
+        learning_rate=args.learning_rate,
+        parameters=model.parameters(),
+        epsilon=args.adam_epsilon,
+        grad_clip=grad_clip,
+        weight_decay=args.weight_decay)
+
+    # Train!
+    logger.info("***** Running training *****")
+    logger.info("  Num examples = {}".format(len(train_dataset)))
+    logger.info("  Num Epochs = {}".format(args.num_train_epochs))
+    logger.info("  Instantaneous batch size per GPU = {}".format(
+        args.per_gpu_train_batch_size))
+    logger.info(
+        "  Total train batch size (w. parallel, distributed & accumulation) = {}".
+        format(args.train_batch_size * paddle.distributed.get_world_size()))
+    logger.info("  Total optimization steps = {}".format(t_total))
+
+    global_step = 0
+    model.clear_gradients()
+    train_dataloader_len = len(train_dataloader)
+    best_metirc = {'f1': 0}
+    model.train()
+
+    for epoch in range(int(args.num_train_epochs)):
+        for step, batch in enumerate(train_dataloader):
+            outputs = model(**batch)
+            # model outputs are always tuple in ppnlp (see doc)
+            loss = outputs['loss']
+            loss = loss.mean()
+
+            logger.info(
+                "epoch: [{}/{}], iter: [{}/{}], global_step:{}, train loss: {}, lr: {}".
+                format(epoch, args.num_train_epochs, step, train_dataloader_len,
+                       global_step, np.mean(loss.numpy()), optimizer.get_lr()))
+
+            loss.backward()
+            optimizer.step()
+            optimizer.clear_grad()
+            # lr_scheduler.step()  # Update learning rate schedule
+
+            global_step += 1
+
+            if (paddle.distributed.get_rank() == 0 and args.eval_steps > 0 and
+                    global_step % args.eval_steps == 0):
+                # Log metrics
+                if (paddle.distributed.get_rank() == 0 and args.
+                        evaluate_during_training):  # Only evaluate when single GPU otherwise metrics may not average well
+                    results = evaluate(model, eval_dataloader, logger)
+                    if results['f1'] > best_metirc['f1']:
+                        best_metirc = results
+                        output_dir = os.path.join(args.output_dir,
+                                                  "checkpoint-best")
+                        os.makedirs(output_dir, exist_ok=True)
+                        model.save_pretrained(output_dir)
+                        tokenizer.save_pretrained(output_dir)
+                        paddle.save(args,
+                                    os.path.join(output_dir,
+                                                 "training_args.bin"))
+                        logger.info("Saving model checkpoint to {}".format(
+                            output_dir))
+                    logger.info("eval results: {}".format(results))
+                    logger.info("best_metirc: {}".format(best_metirc))
+
+            if (paddle.distributed.get_rank() == 0 and args.save_steps > 0 and
+                    global_step % args.save_steps == 0):
+                # Save model checkpoint
+                output_dir = os.path.join(args.output_dir, "checkpoint-latest")
+                os.makedirs(output_dir, exist_ok=True)
+                if paddle.distributed.get_rank() == 0:
+                    model.save_pretrained(output_dir)
+                    tokenizer.save_pretrained(output_dir)
+                    paddle.save(args,
+                                os.path.join(output_dir, "training_args.bin"))
+                    logger.info("Saving model checkpoint to {}".format(
+                        output_dir))
+    logger.info("best_metirc: {}".format(best_metirc))
+
+
+if __name__ == "__main__":
+    args = parse_args()
+    os.makedirs(args.output_dir, exist_ok=True)
+    train(args)

ppstructure/vqa/train_ser.py

@@ -12,8 +12,13 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import sys
 import os
+import sys
+
+__dir__ = os.path.dirname(os.path.abspath(__file__))
+sys.path.append(__dir__)
+sys.path.append(os.path.abspath(os.path.join(__dir__, '../..')))
+
 import random
 import copy
 import logging
@@ -26,8 +31,9 @@ from paddlenlp.transformers import LayoutXLMModel, LayoutXLMTokenizer, LayoutXLM
 from xfun import XFUNDataset
 from utils import parse_args
 from utils import get_bio_label_maps
+from utils import print_arguments
 
-logger = logging.getLogger(__name__)
+from ppocr.utils.logging import get_logger
 
 
 def set_seed(args):
@@ -38,17 +44,8 @@ def set_seed(args):
 
 def train(args):
     os.makedirs(args.output_dir, exist_ok=True)
-    logging.basicConfig(
-        filename=os.path.join(args.output_dir, "train.log")
-        if paddle.distributed.get_rank() == 0 else None,
-        format="%(asctime)s - %(levelname)s - %(name)s -   %(message)s",
-        datefmt="%m/%d/%Y %H:%M:%S",
-        level=logging.INFO
-        if paddle.distributed.get_rank() == 0 else logging.WARN, )
-
-    ch = logging.StreamHandler()
-    ch.setLevel(logging.DEBUG)
-    logger.addHandler(ch)
+    logger = get_logger(log_file=os.path.join(args.output_dir, "train.log"))
+    print_arguments(args, logger)
 
     label2id_map, id2label_map = get_bio_label_maps(args.label_map_path)
     pad_token_label_id = paddle.nn.CrossEntropyLoss().ignore_index
@@ -136,10 +133,10 @@ def train(args):
             loss = outputs[0]
             loss = loss.mean()
             logger.info(
-                "[epoch {}/{}][iter: {}/{}] lr: {:.5f}, train loss: {:.5f}, ".
+                "epoch: [{}/{}], iter: [{}/{}], global_step:{}, train loss: {}, lr: {}".
                 format(epoch_id, args.num_train_epochs, step,
-                       len(train_dataloader),
-                       lr_scheduler.get_lr(), loss.numpy()[0]))
+                       len(train_dataloader), global_step,
+                       loss.numpy()[0], lr_scheduler.get_lr()))
 
             loss.backward()
             tr_loss += loss.item()
@@ -154,13 +151,9 @@ def train(args):
                 # Only evaluate when single GPU otherwise metrics may not average well
                 if paddle.distributed.get_rank(
                 ) == 0 and args.evaluate_during_training:
-                    results, _ = evaluate(
-                        args,
-                        model,
-                        tokenizer,
-                        label2id_map,
-                        id2label_map,
-                        pad_token_label_id, )
+                    results, _ = evaluate(args, model, tokenizer, label2id_map,
+                                          id2label_map, pad_token_label_id,
+                                          logger)
 
                     if best_metrics is None or results["f1"] >= best_metrics[
                             "f1"]:
@@ -204,6 +197,7 @@ def evaluate(args,
              label2id_map,
              id2label_map,
              pad_token_label_id,
+             logger,
              prefix=""):
     eval_dataset = XFUNDataset(
         tokenizer,
@@ -299,15 +293,6 @@ def evaluate(args,
     return results, preds_list
 
 
-def print_arguments(args):
-    """print arguments"""
-    print('-----------  Configuration Arguments -----------')
-    for arg, value in sorted(vars(args).items()):
-        print('%s: %s' % (arg, value))
-    print('------------------------------------------------')
-
-
 if __name__ == "__main__":
     args = parse_args()
-    print_arguments(args)
     train(args)

ppstructure/vqa/utils.py

@@ -24,8 +24,6 @@ import paddle
 
 from PIL import Image, ImageDraw, ImageFont
 
-from paddleocr import PaddleOCR
-
 
 def get_bio_label_maps(label_map_path):
     with open(label_map_path, "r") as fin:
@@ -66,9 +64,9 @@ def get_image_file_list(img_file):
 
 def draw_ser_results(image,
                      ocr_results,
-                     font_path="../doc/fonts/simfang.ttf",
+                     font_path="../../doc/fonts/simfang.ttf",
                      font_size=18):
-    np.random.seed(0)
+    np.random.seed(2021)
     color = (np.random.permutation(range(255)),
              np.random.permutation(range(255)),
              np.random.permutation(range(255)))
@@ -82,38 +80,64 @@ def draw_ser_results(image,
     draw = ImageDraw.Draw(img_new)
 
     font = ImageFont.truetype(font_path, font_size, encoding="utf-8")
-
     for ocr_info in ocr_results:
         if ocr_info["pred_id"] not in color_map:
             continue
         color = color_map[ocr_info["pred_id"]]
-
-        # draw ocr results outline
-        bbox = ocr_info["bbox"]
-        bbox = ((bbox[0], bbox[1]), (bbox[2], bbox[3]))
-        draw.rectangle(bbox, fill=color)
-
-        # draw ocr results
         text = "{}: {}".format(ocr_info["pred"], ocr_info["text"])
-        start_y = max(0, bbox[0][1] - font_size)
-        tw = font.getsize(text)[0]
-        draw.rectangle(
-            [(bbox[0][0] + 1, start_y), (bbox[0][0] + tw + 1,
-                                         start_y + font_size)],
-            fill=(0, 0, 255))
-        draw.text(
-            (bbox[0][0] + 1, start_y), text, fill=(255, 255, 255), font=font)
+
+        draw_box_txt(ocr_info["bbox"], text, draw, font, font_size, color)
 
     img_new = Image.blend(image, img_new, 0.5)
     return np.array(img_new)
 
 
-def build_ocr_engine(rec_model_dir, det_model_dir):
-    ocr_engine = PaddleOCR(
-        rec_model_dir=rec_model_dir,
-        det_model_dir=det_model_dir,
-        use_angle_cls=False)
-    return ocr_engine
+def draw_box_txt(bbox, text, draw, font, font_size, color):
+    # draw ocr results outline
+    bbox = ((bbox[0], bbox[1]), (bbox[2], bbox[3]))
+    draw.rectangle(bbox, fill=color)
+
+    # draw ocr results
+    start_y = max(0, bbox[0][1] - font_size)
+    tw = font.getsize(text)[0]
+    draw.rectangle(
+        [(bbox[0][0] + 1, start_y), (bbox[0][0] + tw + 1, start_y + font_size)],
+        fill=(0, 0, 255))
+    draw.text((bbox[0][0] + 1, start_y), text, fill=(255, 255, 255), font=font)
+
+
+def draw_re_results(image,
+                    result,
+                    font_path="../../doc/fonts/simfang.ttf",
+                    font_size=18):
+    np.random.seed(0)
+    if isinstance(image, np.ndarray):
+        image = Image.fromarray(image)
+    img_new = image.copy()
+    draw = ImageDraw.Draw(img_new)
+
+    font = ImageFont.truetype(font_path, font_size, encoding="utf-8")
+    color_head = (0, 0, 255)
+    color_tail = (255, 0, 0)
+    color_line = (0, 255, 0)
+
+    for ocr_info_head, ocr_info_tail in result:
+        draw_box_txt(ocr_info_head["bbox"], ocr_info_head["text"], draw, font,
+                     font_size, color_head)
+        draw_box_txt(ocr_info_tail["bbox"], ocr_info_tail["text"], draw, font,
+                     font_size, color_tail)
+
+        center_head = (
+            (ocr_info_head['bbox'][0] + ocr_info_head['bbox'][2]) // 2,
+            (ocr_info_head['bbox'][1] + ocr_info_head['bbox'][3]) // 2)
+        center_tail = (
+            (ocr_info_tail['bbox'][0] + ocr_info_tail['bbox'][2]) // 2,
+            (ocr_info_tail['bbox'][1] + ocr_info_tail['bbox'][3]) // 2)
+
+        draw.line([center_head, center_tail], fill=color_line, width=5)
+
+    img_new = Image.blend(image, img_new, 0.5)
+    return np.array(img_new)
 
 
 # pad sentences
@@ -130,7 +154,7 @@ def pad_sentences(tokenizer,
         len(encoded_inputs["input_ids"]) // max_seq_len + 1) * max_seq_len
 
     needs_to_be_padded = pad_to_max_seq_len and \
-                         max_seq_len and len(encoded_inputs["input_ids"]) < max_seq_len
+        max_seq_len and len(encoded_inputs["input_ids"]) < max_seq_len
 
     if needs_to_be_padded:
         difference = max_seq_len - len(encoded_inputs["input_ids"])
@@ -162,6 +186,9 @@ def split_page(encoded_inputs, max_seq_len=512):
     truncate is often used in training process
     """
     for key in encoded_inputs:
+        if key == 'entities':
+            encoded_inputs[key] = [encoded_inputs[key]]
+            continue
         encoded_inputs[key] = paddle.to_tensor(encoded_inputs[key])
         if encoded_inputs[key].ndim <= 1:  # for input_ids, att_mask and so on
             encoded_inputs[key] = encoded_inputs[key].reshape([-1, max_seq_len])
@@ -184,14 +211,14 @@ def preprocess(
     height = ori_img.shape[0]
     width = ori_img.shape[1]
 
-    img = cv2.resize(ori_img,
-                     (224, 224)).transpose([2, 0, 1]).astype(np.float32)
+    img = cv2.resize(ori_img, img_size).transpose([2, 0, 1]).astype(np.float32)
 
     segment_offset_id = []
     words_list = []
     bbox_list = []
     input_ids_list = []
     token_type_ids_list = []
+    entities = []
 
     for info in ocr_info:
         # x1, y1, x2, y2
@@ -211,6 +238,13 @@ def preprocess(
             encode_res["token_type_ids"] = encode_res["token_type_ids"][1:-1]
             encode_res["attention_mask"] = encode_res["attention_mask"][1:-1]
 
+        # for re
+        entities.append({
+            "start": len(input_ids_list),
+            "end": len(input_ids_list) + len(encode_res["input_ids"]),
+            "label": "O",
+        })
+
         input_ids_list.extend(encode_res["input_ids"])
         token_type_ids_list.extend(encode_res["token_type_ids"])
         bbox_list.extend([bbox] * len(encode_res["input_ids"]))
@@ -222,6 +256,7 @@ def preprocess(
         "token_type_ids": token_type_ids_list,
         "bbox": bbox_list,
         "attention_mask": [1] * len(input_ids_list),
+        "entities": entities
     }
 
     encoded_inputs = pad_sentences(
@@ -294,35 +329,64 @@ def merge_preds_list_with_ocr_info(ocr_info, segment_offset_id, preds_list,
     return ocr_info
 
 
+def print_arguments(args, logger=None):
+    print_func = logger.info if logger is not None else print
+    """print arguments"""
+    print_func('-----------  Configuration Arguments -----------')
+    for arg, value in sorted(vars(args).items()):
+        print_func('%s: %s' % (arg, value))
+    print_func('------------------------------------------------')
+
+
 def parse_args():
     parser = argparse.ArgumentParser()
     # Required parameters
     # yapf: disable
-    parser.add_argument("--model_name_or_path", default=None, type=str, required=True,)
-    parser.add_argument("--train_data_dir", default=None, type=str, required=False,)
-    parser.add_argument("--train_label_path", default=None, type=str, required=False,)
-    parser.add_argument("--eval_data_dir", default=None, type=str, required=False,)
-    parser.add_argument("--eval_label_path", default=None, type=str, required=False,)
+    parser.add_argument("--model_name_or_path",
+                        default=None, type=str, required=True,)
+    parser.add_argument("--re_model_name_or_path",
+                        default=None, type=str, required=False,)
+    parser.add_argument("--train_data_dir", default=None,
+                        type=str, required=False,)
+    parser.add_argument("--train_label_path", default=None,
+                        type=str, required=False,)
+    parser.add_argument("--eval_data_dir", default=None,
+                        type=str, required=False,)
+    parser.add_argument("--eval_label_path", default=None,
+                        type=str, required=False,)
     parser.add_argument("--output_dir", default=None, type=str, required=True,)
     parser.add_argument("--max_seq_length", default=512, type=int,)
     parser.add_argument("--evaluate_during_training", action="store_true",)
-    parser.add_argument("--per_gpu_train_batch_size", default=8, type=int, help="Batch size per GPU/CPU for training.",)
-    parser.add_argument("--per_gpu_eval_batch_size", default=8, type=int, help="Batch size per GPU/CPU for eval.",)
-    parser.add_argument("--learning_rate", default=5e-5, type=float, help="The initial learning rate for Adam.",)
-    parser.add_argument("--weight_decay", default=0.0, type=float, help="Weight decay if we apply some.",)
-    parser.add_argument("--adam_epsilon", default=1e-8, type=float, help="Epsilon for Adam optimizer.",)
-    parser.add_argument("--max_grad_norm", default=1.0, type=float, help="Max gradient norm.",)
-    parser.add_argument("--num_train_epochs", default=3, type=int, help="Total number of training epochs to perform.",)
-    parser.add_argument("--warmup_steps", default=0, type=int, help="Linear warmup over warmup_steps.",)
-    parser.add_argument("--eval_steps", type=int, default=10, help="eval every X updates steps.",)
-    parser.add_argument("--save_steps", type=int, default=50, help="Save checkpoint every X updates steps.",)
-    parser.add_argument("--seed", type=int, default=2048, help="random seed for initialization",)
+    parser.add_argument("--per_gpu_train_batch_size", default=8,
+                        type=int, help="Batch size per GPU/CPU for training.",)
+    parser.add_argument("--per_gpu_eval_batch_size", default=8,
+                        type=int, help="Batch size per GPU/CPU for eval.",)
+    parser.add_argument("--learning_rate", default=5e-5,
+                        type=float, help="The initial learning rate for Adam.",)
+    parser.add_argument("--weight_decay", default=0.0,
+                        type=float, help="Weight decay if we apply some.",)
+    parser.add_argument("--adam_epsilon", default=1e-8,
+                        type=float, help="Epsilon for Adam optimizer.",)
+    parser.add_argument("--max_grad_norm", default=1.0,
+                        type=float, help="Max gradient norm.",)
+    parser.add_argument("--num_train_epochs", default=3, type=int,
+                        help="Total number of training epochs to perform.",)
+    parser.add_argument("--warmup_steps", default=0, type=int,
+                        help="Linear warmup over warmup_steps.",)
+    parser.add_argument("--eval_steps", type=int, default=10,
+                        help="eval every X updates steps.",)
+    parser.add_argument("--save_steps", type=int, default=50,
+                        help="Save checkpoint every X updates steps.",)
+    parser.add_argument("--seed", type=int, default=2048,
+                        help="random seed for initialization",)
 
     parser.add_argument("--ocr_rec_model_dir", default=None, type=str, )
     parser.add_argument("--ocr_det_model_dir", default=None, type=str, )
-    parser.add_argument("--label_map_path", default="./labels/labels_ser.txt", type=str, required=False, )
+    parser.add_argument(
+        "--label_map_path", default="./labels/labels_ser.txt", type=str, required=False, )
     parser.add_argument("--infer_imgs", default=None, type=str, required=False)
-    parser.add_argument("--ocr_json_path", default=None, type=str, required=False, help="ocr prediction results")
+    parser.add_argument("--ocr_json_path", default=None,
+                        type=str, required=False, help="ocr prediction results")
     # yapf: enable
     args = parser.parse_args()
     return args