| 
									
										
										
										
											2022-04-18 07:28:22 +00:00
										 |  |  |  | # 基于Python预测引擎推理
 | 
					
						
							|  |  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2022-10-10 05:39:41 +00:00
										 |  |  |  | - [1. 版面信息抽取](#1-版面信息抽取) | 
					
						
							|  |  |  |  |   - [1.1 版面分析+表格识别](#11-版面分析表格识别) | 
					
						
							|  |  |  |  |   - [1.2 版面分析](#12-版面分析) | 
					
						
							|  |  |  |  |   - [1.3 表格识别](#13-表格识别) | 
					
						
							|  |  |  |  | - [2. 关键信息抽取](#2-关键信息抽取) | 
					
						
							|  |  |  |  |   - [2.1 SER](#21-ser) | 
					
						
							|  |  |  |  |   - [2.2 RE+SER](#22-reser) | 
					
						
							| 
									
										
										
										
											2022-04-18 07:28:22 +00:00
										 |  |  |  | 
 | 
					
						
							|  |  |  |  | <a name="1"></a> | 
					
						
							| 
									
										
										
										
											2022-08-22 13:38:31 +00:00
										 |  |  |  | ## 1. 版面信息抽取
 | 
					
						
							| 
									
										
										
										
											2022-04-22 13:24:45 +08:00
										 |  |  |  | 进入`ppstructure`目录 | 
					
						
							| 
									
										
										
										
											2022-04-18 07:28:22 +00:00
										 |  |  |  | 
 | 
					
						
							|  |  |  |  | ```bash | 
					
						
							|  |  |  |  | cd ppstructure | 
					
						
							| 
									
										
										
										
											2022-08-22 13:38:31 +00:00
										 |  |  |  | ``` | 
					
						
							| 
									
										
										
										
											2022-04-22 13:24:45 +08:00
										 |  |  |  | 下载模型 | 
					
						
							|  |  |  |  | ```bash | 
					
						
							| 
									
										
										
										
											2022-04-18 07:28:22 +00:00
										 |  |  |  | mkdir inference && cd inference | 
					
						
							| 
									
										
										
										
											2022-10-12 14:52:33 +08:00
										 |  |  |  | # 下载PP-StructureV2版面分析模型并解压
 | 
					
						
							| 
									
										
										
										
											2022-08-22 06:58:09 +00:00
										 |  |  |  | wget https://paddleocr.bj.bcebos.com/ppstructure/models/layout/picodet_lcnet_x1_0_layout_infer.tar && tar xf picodet_lcnet_x1_0_layout_infer.tar | 
					
						
							|  |  |  |  | # 下载PP-OCRv3文本检测模型并解压
 | 
					
						
							| 
									
										
										
										
											2022-08-22 08:20:20 +00:00
										 |  |  |  | wget https://paddleocr.bj.bcebos.com/PP-OCRv3/chinese/ch_PP-OCRv3_det_infer.tar && tar xf ch_PP-OCRv3_det_infer.tar | 
					
						
							| 
									
										
										
										
											2022-08-22 06:58:09 +00:00
										 |  |  |  | # 下载PP-OCRv3文本识别模型并解压
 | 
					
						
							| 
									
										
										
										
											2022-08-22 08:20:20 +00:00
										 |  |  |  | wget https://paddleocr.bj.bcebos.com/PP-OCRv3/chinese/ch_PP-OCRv3_rec_infer.tar && tar xf ch_PP-OCRv3_rec_infer.tar | 
					
						
							| 
									
										
										
										
											2022-10-12 14:52:33 +08:00
										 |  |  |  | # 下载PP-StructureV2表格识别模型并解压
 | 
					
						
							| 
									
										
										
										
											2022-08-22 06:58:09 +00:00
										 |  |  |  | wget https://paddleocr.bj.bcebos.com/ppstructure/models/slanet/ch_ppstructure_mobile_v2.0_SLANet_infer.tar && tar xf ch_ppstructure_mobile_v2.0_SLANet_infer.tar | 
					
						
							| 
									
										
										
										
											2022-04-18 07:28:22 +00:00
										 |  |  |  | cd .. | 
					
						
							| 
									
										
										
										
											2022-04-22 13:24:45 +08:00
										 |  |  |  | ``` | 
					
						
							|  |  |  |  | <a name="1.1"></a> | 
					
						
							|  |  |  |  | ### 1.1 版面分析+表格识别
 | 
					
						
							|  |  |  |  | ```bash | 
					
						
							| 
									
										
										
										
											2022-08-22 08:20:20 +00:00
										 |  |  |  | python3 predict_system.py --det_model_dir=inference/ch_PP-OCRv3_det_infer \ | 
					
						
							|  |  |  |  |                           --rec_model_dir=inference/ch_PP-OCRv3_rec_infer \ | 
					
						
							| 
									
										
										
										
											2022-08-22 06:58:09 +00:00
										 |  |  |  |                           --table_model_dir=inference/ch_ppstructure_mobile_v2.0_SLANet_infer \ | 
					
						
							|  |  |  |  |                           --layout_model_dir=inference/picodet_lcnet_x1_0_layout_infer \ | 
					
						
							| 
									
										
										
										
											2022-04-22 13:24:45 +08:00
										 |  |  |  |                           --image_dir=./docs/table/1.png \ | 
					
						
							| 
									
										
										
										
											2022-04-18 07:28:22 +00:00
										 |  |  |  |                           --rec_char_dict_path=../ppocr/utils/ppocr_keys_v1.txt \ | 
					
						
							| 
									
										
										
										
											2022-08-22 06:58:09 +00:00
										 |  |  |  |                           --table_char_dict_path=../ppocr/utils/dict/table_structure_dict_ch.txt \ | 
					
						
							| 
									
										
										
										
											2022-04-22 13:24:45 +08:00
										 |  |  |  |                           --output=../output \ | 
					
						
							| 
									
										
										
										
											2022-04-18 07:28:22 +00:00
										 |  |  |  |                           --vis_font_path=../doc/fonts/simfang.ttf | 
					
						
							|  |  |  |  | ``` | 
					
						
							| 
									
										
										
										
											2022-04-22 13:24:45 +08:00
										 |  |  |  | 运行完成后,每张图片会在`output`字段指定的目录下的`structure`目录下有一个同名目录,图片里的每个表格会存储为一个excel,图片区域会被裁剪之后保存下来,excel文件和图片名为表格在图片里的坐标。详细的结果会存储在`res.txt`文件中。 | 
					
						
							|  |  |  |  | 
 | 
					
						
							|  |  |  |  | <a name="1.2"></a> | 
					
						
							|  |  |  |  | ### 1.2 版面分析
 | 
					
						
							|  |  |  |  | ```bash | 
					
						
							| 
									
										
										
										
											2022-08-22 06:58:09 +00:00
										 |  |  |  | python3 predict_system.py --layout_model_dir=inference/picodet_lcnet_x1_0_layout_infer \ | 
					
						
							|  |  |  |  |                           --image_dir=./docs/table/1.png \ | 
					
						
							|  |  |  |  |                           --output=../output \ | 
					
						
							|  |  |  |  |                           --table=false \ | 
					
						
							|  |  |  |  |                           --ocr=false | 
					
						
							| 
									
										
										
										
											2022-04-22 13:24:45 +08:00
										 |  |  |  | ``` | 
					
						
							|  |  |  |  | 运行完成后,每张图片会在`output`字段指定的目录下的`structure`目录下有一个同名目录,图片区域会被裁剪之后保存下来,图片名为表格在图片里的坐标。版面分析结果会存储在`res.txt`文件中。 | 
					
						
							|  |  |  |  | 
 | 
					
						
							|  |  |  |  | <a name="1.3"></a> | 
					
						
							|  |  |  |  | ### 1.3 表格识别
 | 
					
						
							|  |  |  |  | ```bash | 
					
						
							| 
									
										
										
										
											2022-08-22 08:20:20 +00:00
										 |  |  |  | python3 predict_system.py --det_model_dir=inference/ch_PP-OCRv3_det_infer \ | 
					
						
							|  |  |  |  |                           --rec_model_dir=inference/ch_PP-OCRv3_rec_infer \ | 
					
						
							| 
									
										
										
										
											2022-08-22 06:58:09 +00:00
										 |  |  |  |                           --table_model_dir=inference/ch_ppstructure_mobile_v2.0_SLANet_infer \ | 
					
						
							| 
									
										
										
										
											2022-04-22 13:24:45 +08:00
										 |  |  |  |                           --image_dir=./docs/table/table.jpg \ | 
					
						
							|  |  |  |  |                           --rec_char_dict_path=../ppocr/utils/ppocr_keys_v1.txt \ | 
					
						
							| 
									
										
										
										
											2022-08-22 06:58:09 +00:00
										 |  |  |  |                           --table_char_dict_path=../ppocr/utils/dict/table_structure_dict_ch.txt \ | 
					
						
							| 
									
										
										
										
											2022-04-22 13:24:45 +08:00
										 |  |  |  |                           --output=../output \ | 
					
						
							|  |  |  |  |                           --vis_font_path=../doc/fonts/simfang.ttf \ | 
					
						
							|  |  |  |  |                           --layout=false | 
					
						
							|  |  |  |  | ``` | 
					
						
							|  |  |  |  | 运行完成后,每张图片会在`output`字段指定的目录下的`structure`目录下有一个同名目录,表格会存储为一个excel,excel文件名为`[0,0,img_h,img_w]`。 | 
					
						
							| 
									
										
										
										
											2022-04-18 07:28:22 +00:00
										 |  |  |  | 
 | 
					
						
							|  |  |  |  | <a name="2"></a> | 
					
						
							| 
									
										
										
										
											2022-08-22 09:52:23 +08:00
										 |  |  |  | ## 2. 关键信息抽取
 | 
					
						
							| 
									
										
										
										
											2022-04-18 07:28:22 +00:00
										 |  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2022-10-10 05:39:41 +00:00
										 |  |  |  | ### 2.1 SER
 | 
					
						
							|  |  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2022-04-18 07:28:22 +00:00
										 |  |  |  | ```bash | 
					
						
							|  |  |  |  | cd ppstructure | 
					
						
							|  |  |  |  | 
 | 
					
						
							|  |  |  |  | mkdir inference && cd inference | 
					
						
							| 
									
										
										
										
											2022-08-22 09:52:23 +08:00
										 |  |  |  | # 下载SER XFUND 模型并解压
 | 
					
						
							|  |  |  |  | wget https://paddleocr.bj.bcebos.com/ppstructure/models/vi_layoutxlm/ser_vi_layoutxlm_xfund_infer.tar && tar -xf ser_vi_layoutxlm_xfund_infer.tar | 
					
						
							| 
									
										
										
										
											2022-04-18 07:28:22 +00:00
										 |  |  |  | cd .. | 
					
						
							| 
									
										
										
										
											2022-10-10 06:31:44 +00:00
										 |  |  |  | python3 predict_system.py \ | 
					
						
							| 
									
										
										
										
											2022-08-22 09:52:23 +08:00
										 |  |  |  |   --kie_algorithm=LayoutXLM \ | 
					
						
							| 
									
										
										
										
											2022-10-10 06:31:44 +00:00
										 |  |  |  |   --ser_model_dir=./inference/ser_vi_layoutxlm_xfund_infer \ | 
					
						
							| 
									
										
										
										
											2022-08-22 09:52:23 +08:00
										 |  |  |  |   --image_dir=./docs/kie/input/zh_val_42.jpg \ | 
					
						
							|  |  |  |  |   --ser_dict_path=../ppocr/utils/dict/kie_dict/xfund_class_list.txt \ | 
					
						
							|  |  |  |  |   --vis_font_path=../doc/fonts/simfang.ttf \ | 
					
						
							| 
									
										
										
										
											2022-10-10 06:31:44 +00:00
										 |  |  |  |   --ocr_order_method="tb-yx" \ | 
					
						
							|  |  |  |  |   --mode=kie | 
					
						
							| 
									
										
										
										
											2022-04-18 07:28:22 +00:00
										 |  |  |  | ``` | 
					
						
							| 
									
										
										
										
											2022-08-22 09:52:23 +08:00
										 |  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2022-08-21 10:55:49 +08:00
										 |  |  |  | 运行完成后,每张图片会在`output`字段指定的目录下的`kie`目录下存放可视化之后的图片,图片名和输入图片名一致。 | 
					
						
							| 
									
										
										
										
											2022-10-10 05:39:41 +00:00
										 |  |  |  | 
 | 
					
						
							|  |  |  |  | ### 2.2 RE+SER
 | 
					
						
							|  |  |  |  | 
 | 
					
						
							|  |  |  |  | ```bash | 
					
						
							|  |  |  |  | cd ppstructure | 
					
						
							|  |  |  |  | 
 | 
					
						
							|  |  |  |  | mkdir inference && cd inference | 
					
						
							|  |  |  |  | # 下载RE SER XFUND 模型并解压
 | 
					
						
							|  |  |  |  | wget https://paddleocr.bj.bcebos.com/ppstructure/models/vi_layoutxlm/ser_vi_layoutxlm_xfund_infer.tar && tar -xf ser_vi_layoutxlm_xfund_infer.tar | 
					
						
							|  |  |  |  | wget https://paddleocr.bj.bcebos.com/ppstructure/models/vi_layoutxlm/re_vi_layoutxlm_xfund_infer.tar && tar -xf re_vi_layoutxlm_xfund_infer.tar | 
					
						
							|  |  |  |  | cd .. | 
					
						
							|  |  |  |  | 
 | 
					
						
							|  |  |  |  | python3 predict_system.py \ | 
					
						
							|  |  |  |  |   --kie_algorithm=LayoutXLM \ | 
					
						
							|  |  |  |  |   --re_model_dir=./inference/re_vi_layoutxlm_xfund_infer \ | 
					
						
							|  |  |  |  |   --ser_model_dir=./inference/ser_vi_layoutxlm_xfund_infer \ | 
					
						
							|  |  |  |  |   --image_dir=./docs/kie/input/zh_val_42.jpg \ | 
					
						
							|  |  |  |  |   --ser_dict_path=../ppocr/utils/dict/kie_dict/xfund_class_list.txt \ | 
					
						
							|  |  |  |  |   --vis_font_path=../doc/fonts/simfang.ttf \ | 
					
						
							|  |  |  |  |   --ocr_order_method="tb-yx" \ | 
					
						
							|  |  |  |  |   --mode=kie | 
					
						
							|  |  |  |  | ``` | 
					
						
							|  |  |  |  | 
 | 
					
						
							|  |  |  |  | 运行完成后,每张图片会在`output`字段指定的目录下的`kie`目录下有一个同名目录,目录中存放可视化图片和预测结果。 |