-
|
Beta Was this translation helpful? Give feedback.
Replies: 3 comments 4 replies
-
|
split_idxs = np.argwhere(abs(minus_res) > thresh).squeeze() 输出变量是一个数组，可能是空数组，但仍然是一个数组； |
Beta Was this translation helpful? Give feedback.
-
|
能否留一下出现这个异常的图片呢？ |
Beta Was this translation helpful? Give feedback.
-
|
图片我找到了，但是目前我的开发环境与网络环境是隔离，我无法将图片上传给你。我可以描述一下：是pdf的目录页，但是底部多了一个框，里面是个签名。算法应该是在英文这个框，判断使用wired方案。 |
Beta Was this translation helpful? Give feedback.
-
|
非常感谢指出这个问题，我修一下这里 |
Beta Was this translation helpful? Give feedback.
-
|
但是这里为空，其实相当于完全失去了表格识别的功能，提取目录的话，建议直接进行ocr识别会更好，担心排版问题的话，可以根据ocr框的间距来处理文字间距 |
Beta Was this translation helpful? Give feedback.
-
|
OK，我帮你测试一下
…---Original---
From: ***@***.***>
Date: Fri, Sep 27, 2024 14:56 PM
To: ***@***.***>;
Cc: "Kyrie ***@***.******@***.***>;
Subject: Re: [RapidAI/TableStructureRec] bug :wired_table_rec/table_recover.py(Discussion #33)
能否留一下出现这个异常的图片呢？
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you authored the thread.Message ID: ***@***.***>
|
Beta Was this translation helpful? Give feedback.
-
|
其实这里对我的功能不太影响，我的场景是处理pdf文档，里面含有大量的的表格！我没有单独对某一页进行针对性的处理！只是偶然遇到这个问题，交流一下，谢谢
…---Original---
From: ***@***.***>
Date: Fri, Sep 27, 2024 15:19 PM
To: ***@***.***>;
Cc: "Kyrie ***@***.******@***.***>;
Subject: Re: [RapidAI/TableStructureRec] bug :wired_table_rec/table_recover.py(Discussion #33)
但是这里为空，其实相当于完全失去了表格识别的功能，提取目录的话，建议直接进行ocr识别会更好，担心排版问题的话，可以根据ocr框的间距来处理文字间距
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you authored the thread.Message ID: ***@***.***>
|
Beta Was this translation helpful? Give feedback.
split_idxs = np.argwhere(abs(minus_res) > thresh).squeeze() 输出变量是一个数组，可能是空数组，但仍然是一个数组；
if split_idxs.ndim == 0: 这个条件永远不会为真，max(split_idxs) 就会报错。