-
Notifications
You must be signed in to change notification settings - Fork 388
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
txt文件与图像间的关系 #354
Comments
我写这个,是为了给那些,下不到 synth90k数据 ,但是,又想自己做 数据,却又不明白 txt 含义的朋友一个 友好的 提示。这个,不算是问题,只是对 txt 文件的一个解说。 |
@wandaoyi Thanks a lot. I will leave it here:) |
You are welcome |
Thanks!This is a very useful information. |
you are welcome
…------------------ 原始邮件 ------------------
发件人: "York Law"<[email protected]>;
发送时间: 2019年11月8日(星期五) 晚上7:55
收件人: "MaybeShewill-CV/CRNN_Tensorflow"<[email protected]>;
抄送: "求道"<[email protected]>; "Mention"<[email protected]>;
主题: Re: [MaybeShewill-CV/CRNN_Tensorflow] txt文件与图像间的关系 (#354)
Thanks!This is a very useful information.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or unsubscribe.
|
中文怎么办?这是以词来标注的吧,中文序列呢请问 |
@LW-CVer 数据里面很多都不是词组而是句子 |
中文是以 词 来标序号的。你仔细阅读以下,我记得,之前是有说明白的。只是,以前我无法上传图片,没办法图文结合讲解。但是,我是有文字说明的
…------------------ 原始邮件 ------------------
发件人: "LW-CVer"<[email protected]>;
发送时间: 2020年3月20日(星期五) 中午11:57
收件人: "MaybeShewill-CV/CRNN_Tensorflow"<[email protected]>;
抄送: "求道"<[email protected]>; "Mention"<[email protected]>;
主题: Re: [MaybeShewill-CV/CRNN_Tensorflow] txt文件与图像间的关系 (#354)
中文怎么办?这是以词来标注的吧,中文序列呢请问
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or unsubscribe.
|
你没发现,项目里面,有一个字典吗?你缕缕项目的思路先吧。
…------------------ 原始邮件 ------------------
发件人: "MaybeShewill-CV"<[email protected]>;
发送时间: 2020年3月20日(星期五) 中午12:52
收件人: "MaybeShewill-CV/CRNN_Tensorflow"<[email protected]>;
抄送: "求道"<[email protected]>; "Mention"<[email protected]>;
主题: Re: [MaybeShewill-CV/CRNN_Tensorflow] txt文件与图像间的关系 (#354)
@LW-CVer 数据里面很多都不是词组而是句子
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or unsubscribe.
|
你好,我也想问中文数据集的图片都是句子,那么lexicon.txt里面也要写句子吗,还是有其他的形式呢 |
这个,随便看一下逻辑,就知道了的呀。我们训练数据,有图片,有label。你的label 怎么来?
然后,将 label 根据 id,转为 数字。
而图片用卷积提取特征,进行损失计算。
这不是思路很简单嘛?
…------------------ 原始邮件 ------------------
发件人: "18796956836"<[email protected]>;
发送时间: 2020年3月31日(星期二) 下午5:05
收件人: "MaybeShewill-CV/CRNN_Tensorflow"<[email protected]>;
抄送: "求道"<[email protected]>; "Mention"<[email protected]>;
主题: Re: [MaybeShewill-CV/CRNN_Tensorflow] txt文件与图像间的关系 (#354)
你好,我也想问中文数据集的图片都是句子,那么lexicon.txt里面也要写句子吗,还是有其他的形式呢
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or unsubscribe.
|
对的。麻烦您了。我就是想问一下,中文数据集lexicon.txt里面写啥,不好意思,给你添麻烦了。 |
您好可以加一下联系方式吗,我想请教您问题,我的qq是992140698
…------------------ 原始邮件 ------------------
发件人: "wandaoyi"<[email protected]>;
发送时间: 2020年3月31日(星期二) 下午5:26
收件人: "MaybeShewill-CV/CRNN_Tensorflow"<[email protected]>;
抄送: "gsdfgdf"<[email protected]>;"Comment"<[email protected]>;
主题: Re: [MaybeShewill-CV/CRNN_Tensorflow] txt文件与图像间的关系 (#354)
这个,随便看一下逻辑,就知道了的呀。我们训练数据,有图片,有label。你的label 怎么来?
然后,将 label 根据 id,转为 数字。
而图片用卷积提取特征,进行损失计算。
这不是思路很简单嘛?
------------------&nbsp;原始邮件&nbsp;------------------
发件人: "18796956836"<[email protected]&gt;;
发送时间: 2020年3月31日(星期二) 下午5:05
收件人: "MaybeShewill-CV/CRNN_Tensorflow"<[email protected]&gt;;
抄送: "求道"<[email protected]&gt;; "Mention"<[email protected]&gt;;
主题: Re: [MaybeShewill-CV/CRNN_Tensorflow] txt文件与图像间的关系 (#354)
你好,我也想问中文数据集的图片都是句子,那么lexicon.txt里面也要写句子吗,还是有其他的形式呢
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or unsubscribe.
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub, or unsubscribe.
|
我都好久没看这个了,你自己捋捋代码和思路,就知道的了。我之前没办法上传图片上去说明(好像是不给传),久了,就忘记了。
…------------------ 原始邮件 ------------------
发件人: "fagrgarhtshtj"<[email protected]>;
发送时间: 2020年3月31日(星期二) 下午5:33
收件人: "MaybeShewill-CV/CRNN_Tensorflow"<[email protected]>;
抄送: "求道"<[email protected]>; "Mention"<[email protected]>;
主题: Re: [MaybeShewill-CV/CRNN_Tensorflow] txt文件与图像间的关系 (#354)
对的。麻烦您了。我就是想问一下,中文数据集lexicon.txt里面写啥,不好意思,给你添麻烦了。
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or unsubscribe.
|
好的,十分感谢。 |
我晚上有空,找找看,如果能找到这个项目,再给你发一份吧我记得,我没用原来的,我是修改了的
…------------------ 原始邮件 ------------------
发件人: "fagrgarhtshtj"<[email protected]>;
发送时间: 2020年3月31日(星期二) 下午5:40
收件人: "MaybeShewill-CV/CRNN_Tensorflow"<[email protected]>;
抄送: "求道"<[email protected]>; "Mention"<[email protected]>;
主题: Re: [MaybeShewill-CV/CRNN_Tensorflow] txt文件与图像间的关系 (#354)
好的,十分感谢。
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or unsubscribe.
|
好的,谢谢
…------------------ 原始邮件 ------------------
发件人: "wandaoyi"<[email protected]>;
发送时间: 2020年3月31日(星期二) 下午5:42
收件人: "MaybeShewill-CV/CRNN_Tensorflow"<[email protected]>;
抄送: "gsdfgdf"<[email protected]>;"Comment"<[email protected]>;
主题: Re: [MaybeShewill-CV/CRNN_Tensorflow] txt文件与图像间的关系 (#354)
我晚上有空,找找看,如果能找到这个项目,再给你发一份吧我记得,我没用原来的,我是修改了的
------------------&nbsp;原始邮件&nbsp;------------------
发件人: "fagrgarhtshtj"<[email protected]&gt;;
发送时间: 2020年3月31日(星期二) 下午5:40
收件人: "MaybeShewill-CV/CRNN_Tensorflow"<[email protected]&gt;;
抄送: "求道"<[email protected]&gt;; "Mention"<[email protected]&gt;;
主题: Re: [MaybeShewill-CV/CRNN_Tensorflow] txt文件与图像间的关系 (#354)
好的,十分感谢。
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or unsubscribe.
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub, or unsubscribe.
|
我没空去给你捋这东西呀,只能你自己去捋了。不然,你让那个作者给你提供一份吧
…------------------ 原始邮件 ------------------
发件人: "fagrgarhtshtj"<[email protected]>;
发送时间: 2020年3月31日(星期二) 下午5:35
收件人: "MaybeShewill-CV/CRNN_Tensorflow"<[email protected]>;
抄送: "求道"<[email protected]>; "Mention"<[email protected]>;
主题: Re: [MaybeShewill-CV/CRNN_Tensorflow] txt文件与图像间的关系 (#354)
您好可以加一下联系方式吗,我想请教您问题,我的qq是992140698
------------------&nbsp;原始邮件&nbsp;------------------
发件人:&nbsp;"wandaoyi"<[email protected]&gt;;
发送时间:&nbsp;2020年3月31日(星期二) 下午5:26
收件人:&nbsp;"MaybeShewill-CV/CRNN_Tensorflow"<[email protected]&gt;;
抄送:&nbsp;"gsdfgdf"<[email protected]&gt;;"Comment"<[email protected]&gt;;
主题:&nbsp;Re: [MaybeShewill-CV/CRNN_Tensorflow] txt文件与图像间的关系 (#354)
这个,随便看一下逻辑,就知道了的呀。我们训练数据,有图片,有label。你的label 怎么来?
然后,将 label 根据 id,转为 数字。
而图片用卷积提取特征,进行损失计算。
这不是思路很简单嘛?
------------------&amp;nbsp;原始邮件&amp;nbsp;------------------
发件人: "18796956836"<[email protected]&amp;gt;;
发送时间: 2020年3月31日(星期二) 下午5:05
收件人: "MaybeShewill-CV/CRNN_Tensorflow"<[email protected]&amp;gt;;
抄送: "求道"<[email protected]&amp;gt;; "Mention"<[email protected]&amp;gt;;
主题: Re: [MaybeShewill-CV/CRNN_Tensorflow] txt文件与图像间的关系 (#354)
你好,我也想问中文数据集的图片都是句子,那么lexicon.txt里面也要写句子吗,还是有其他的形式呢
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or unsubscribe.
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub, or unsubscribe.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or unsubscribe.
|
pre-trained model下载不了,有其他地址么? |
您好,你下载不了,是由于没有科学上网吗?你要先确定你可以科学上网,再去下载。
里面的英文model 是很好的,但是,中文model 不怎么样,需要你自己重新训练的。
…------------------ 原始邮件 ------------------
发件人: "lulupanPPY"<[email protected]>;
发送时间: 2020年4月22日(星期三) 中午12:54
收件人: "MaybeShewill-CV/CRNN_Tensorflow"<[email protected]>;
抄送: "求道"<[email protected]>; "Mention"<[email protected]>;
主题: Re: [MaybeShewill-CV/CRNN_Tensorflow] txt文件与图像间的关系 (#354)
pre-trained model下载不了,有其他地址么?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or unsubscribe.
|
@lulupanPPY @wandaoyi 中文模型之前上传错了 我已经重新上传过链接了 暂时没有dropbox以外的链接存放方式 中文模型路径 https://www.dropbox.com/sh/z22xsn4byddalv3/AAAiIxAHJKbqy44M73ow5znSa?dl=0 可以参考这个issue #390 |
大佬,您可以把这个英文 和 中文的 模型,发一下给我,或百度云给我一下链接吗?我现在,也无法科学上网,都打不开这个网址。
…------------------ 原始邮件 ------------------
发件人: "MaybeShewill-CV"<[email protected]>;
发送时间: 2020年4月22日(星期三) 中午1:34
收件人: "MaybeShewill-CV/CRNN_Tensorflow"<[email protected]>;
抄送: "求道"<[email protected]>; "Mention"<[email protected]>;
主题: Re: [MaybeShewill-CV/CRNN_Tensorflow] txt文件与图像间的关系 (#354)
@lulupanPPY @wandaoyi 中文模型之前上传错了 我已经重新上传过链接了 暂时没有dropbox以外的链接存放方式 中文模型路径 https://www.dropbox.com/sh/z22xsn4byddalv3/AAAiIxAHJKbqy44M73ow5znSa?dl=0 可以参考这个issue #390
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or unsubscribe.
|
@wandaoyi 暂时没有办法:) |
Thanks
…------------------ 原始邮件 ------------------
发件人: "MaybeShewill-CV"<[email protected]>;
发送时间: 2020年4月23日(星期四) 晚上8:22
收件人: "MaybeShewill-CV/CRNN_Tensorflow"<[email protected]>;
抄送: "求道"<[email protected]>; "Mention"<[email protected]>;
主题: Re: [MaybeShewill-CV/CRNN_Tensorflow] txt文件与图像间的关系 (#354)
@wandaoyi 暂时没有办法:)
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or unsubscribe.
|
能发一份模型给我嘛,我下载不了,邮箱[email protected],万分感谢! |
链接:https://pan.baidu.com/s/1fmtWIITiF22vsQbIernl7A
提取码:tuzu
…------------------ 原始邮件 ------------------
发件人: "aaferrero"<[email protected]>;
发送时间: 2020年7月10日(星期五) 晚上6:23
收件人: "MaybeShewill-CV/CRNN_Tensorflow"<[email protected]>;
抄送: "求道"<[email protected]>; "Mention"<[email protected]>;
主题: Re: [MaybeShewill-CV/CRNN_Tensorflow] txt文件与图像间的关系 (#354)
能发一份模型给我嘛,我下载不了,邮箱[email protected],万分感谢!
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or unsubscribe.
|
|
------------------ 原始邮件 ------------------
发件人: "aaferrero"<[email protected]>;
发送时间: 2020年7月11日(星期六) 中午11:05
收件人: "MaybeShewill-CV/CRNN_Tensorflow"<[email protected]>;
抄送: "求道"<[email protected]>; "Mention"<[email protected]>;
主题: Re: [MaybeShewill-CV/CRNN_Tensorflow] txt文件与图像间的关系 (#354)
链接:https://pan.baidu.com/s/1fmtWIITiF22vsQbIernl7A 提取码:tuzu
…
------------------ 原始邮件 ------------------ 发件人: "aaferrero"<[email protected]>; 发送时间: 2020年7月10日(星期五) 晚上6:23 收件人: "MaybeShewill-CV/CRNN_Tensorflow"<[email protected]>; 抄送: "求道"<[email protected]>; "Mention"<[email protected]>; 主题: Re: [MaybeShewill-CV/CRNN_Tensorflow] txt文件与图像间的关系 (#354) 能发一份模型给我嘛,我下载不了,邮箱[email protected],万分感谢! — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.
谢谢!
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or unsubscribe.
|
发了两次图片说明,都发不成功图片。txt文件与图像的关系如图:
链接:https://pan.baidu.com/s/1_qOSRn7Pf0XwTbL-yflQFw
提取码:pouw
希望对大家有用。
…------------------ 原始邮件 ------------------
发件人: "aaferrero"<[email protected]>;
发送时间: 2020年7月11日(星期六) 中午11:05
收件人: "MaybeShewill-CV/CRNN_Tensorflow"<[email protected]>;
抄送: "求道"<[email protected]>; "Mention"<[email protected]>;
主题: Re: [MaybeShewill-CV/CRNN_Tensorflow] txt文件与图像间的关系 (#354)
链接:https://pan.baidu.com/s/1fmtWIITiF22vsQbIernl7A 提取码:tuzu
…
------------------ 原始邮件 ------------------ 发件人: "aaferrero"<[email protected]>; 发送时间: 2020年7月10日(星期五) 晚上6:23 收件人: "MaybeShewill-CV/CRNN_Tensorflow"<[email protected]>; 抄送: "求道"<[email protected]>; "Mention"<[email protected]>; 主题: Re: [MaybeShewill-CV/CRNN_Tensorflow] txt文件与图像间的关系 (#354) 能发一份模型给我嘛,我下载不了,邮箱[email protected],万分感谢! — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.
谢谢!
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or unsubscribe.
|
annotation.txt 文件中,由两个部分组成:image_path 和 label_index。如:
./3000/7/169_deliberations_20271.jpg 20271
它们之间以空格键隔开。
可以根据 image_path 找到目标图像,根据 label_index 在 lexicon.txt 文件中找到对应的 label。
如上面的例子,label 就在 lexicon.txt 的 20271 行,内容为: deliberation
同理,annotation_train.txt 里面的内容,也是一样的意义。
The text was updated successfully, but these errors were encountered: