| 123456789101112131415161718192021222324252627282930 |
- from paddlex import create_pipeline
- import time
- from pathlib import Path
- input_path = "./sample_data/300674-母公司现金流量表-扫描.png"
- pipeline_path = "./PP-StructureV3-zhch.yaml"
- pipeline_name = Path(pipeline_path).stem
- output_path = Path(f"./sample_data/single_pipeline_output/{pipeline_name}/")
- pipeline = create_pipeline(pipeline=pipeline_path)
- # For Image
- output = pipeline.predict(
- input=input_path,
- device="gpu", # 或者 "gpu" 如果你有 GPU 支持
- use_doc_orientation_classify=True, # 开启文档方向分类
- use_doc_unwarping=False, # 开启文档去畸变, 效果不佳
- layout_detection_model_name=None, # 如果要禁用版面分析，可以这样设置，或者依赖其默认行为
- use_seal_recognition=True, # 跳过印章识别
- use_chart_recognition=True, # 跳过图表识别
- use_table_recognition=True, # 开启表格识别
- )
- # 可视化结果并保存 json 结果
- for res in output:
- res.print()
- # res.save_to_json(save_path="sample_data/output")
- # res.save_to_markdown(save_path="sample_data/output")
- res.save_all(save_path=output_path.as_posix()) # 保存所有结果到指定路径
|
