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.
|
Closed
|
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.
|
Closed
|
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.
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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: