Skip to content

中文版面检测(Chinese layout detection),yolov8 is used to detect the layout of Chinese document images。

Notifications You must be signed in to change notification settings

jiangnanboy/layout_analysis

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

利用yolov8对中文文档图片进行版面检测

yolov8 is used to detect the layout of Chinese document images

模型下载、训练及推理

本项目根据开源中文版面数据CDLA ,利用yolov8训练两个模型8mpt与8npt,

CDLA是一个中文文档版面分析数据集,面向中文文献类(论文)场景。包含以下10个label:

正文 标题 图片 图片标题 表格 表格标题 页眉 页脚 注释 公式
Text Title Figure Figure caption Table Table caption Header Footer Reference Equation

8mpt模型与8npt模型下载:

链接:https://pan.baidu.com/s/1YakM5AYrakoG9hYN-w7mJw

提取码:j2za

训练:

from ultralytics import YOLO

def train_model():
    # 加载模型
    print('model load。。。')
    model = YOLO("8npt/best.pt")  # 加载模型
    print('model load completed。。。')
    #使用模型
    model.train(data="img-layout.yaml", epochs=300, device=1)# , lr0=0.0001)  # 训练模型
    metrics = model.val()  # 在验证集上评估模型性能

8npt


8mpt


推理:

from ultralytics import YOLO
def infer():
    model = YOLO('8npt/best.pt')
    results = model('img.jpg')
    print(results[0].plot())
    cv2.imwrite('result.png', results[0].plot())


contact

1、github:https://github.com/jiangnanboy

2、博客:https://www.cnblogs.com/little-horse/

3、邮件:2229029156@qq.com

reference

https://github.com/ultralytics/ultralytics

https://github.com/buptlihang/CDLA

About

中文版面检测(Chinese layout detection),yolov8 is used to detect the layout of Chinese document images。

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages