microsoft

/

layoutlmv3-base-chinese

like 65

Follow

Microsoft 4,772

数据集（dataset）：mmhzlrj/Genealogy
from datasets import load_dataset
dataset = load_dataset("mmhzlrj/Genealogy")

您好！我是一位AI的初学者，了解到layoutlmv3是处理NLP的一个非常强大的多模态大模型，希望使用它做一件非常有意义的事情。但是我不懂如何使用这个模型微调和识别图片来完成我想要实现的族谱修复整理任务：

1. 识别族谱的文字排版，将扫描版的图片转化成可以选择文字的PDF
2. 识别内容，生成以人物卡（姓名：生-死，藏地，学历，子嗣，事迹等一切识别出来的标签内容）
3. 由人物卡连接成树的图形化的家族树

Hello! I am a beginner in AI and have learned that layoutlmv3 is a very powerful multimodal model for handling NLP. I hope to use it to do something very meaningful. But I don't know how to use this model to fine tune and recognize images to complete the Genealogy repair maintenance.

To do list:

1. Recognize the text layout of the genealogy and convert scanned images into PDF with selectable text
2. Identify the content and generate tag content with character cards (name: DOB-DOD, burial site, educational background, descendants, events, etc.)
3. A graphical family tree connected by character cards into a tree

Thanks