summarizing LSTM

[Youcef]

Hi,

In that page https://github.com/tesseract-ocr/tesseract/wiki/VGSLSpecs from officiel github repo, it talks about "summarizing LSTM" and how we can manage to recognize image of both variable height and variable width . Is there a paper, a blog or anything else that provide more explicit information regarding that, that provide explicit architecture examples? I didn't succeed in google it, maybe I'm not using the right keywords. 

Thanks for any help.

Regards

Youcef

[Timothy Snyder]

Do you want to learn more about neural networks or specifically, a "summarizing LSTM" in a neural network?

--
You received this message because you are subscribed to the Google Groups "tesseract-ocr" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tesseract-oc...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/0a940ac0-9000-46c6-9f93-73c78c29068e%40googlegroups.com.

[Purushotham Rao Eravalli]

It will be great if you provide any source where we can get detailed information about the architecture used for tesseract and it's loss functions or so.

Thanks

To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/CABtjQ9KCH-X%3DNLmjdaPMsX9jefMDPpyFD-Sj-XSvWoQURqUvoA%40mail.gmail.com.

[Timothy Snyder]

This page goes into a little more details than the VGSL spec page in the Tesseract repo: https://github.com/mldbai/tensorflow-models/blob/master/street/g3doc/vgslspecs.md

Not specific to Tesseract but this guy's articles have good info on lower-level neural net mechanics in relation to OCR.

To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/CAHrKmrWqvUq%2BukUyR4aDQqzMCQNc9oVcdAUHhcN0CVxEAeeG9Q%40mail.gmail.com.

[Timothy Snyder]

the link for my second sentence ^ https://githubharald.github.io/

[Youcef]

Hi Timothy, I was in fact looking for some more information on how LTSM can deal with two variable dimensions. Thanks for your first link, it looks more detailed regarding that subject.

Le vendredi 6 septembre 2019 15:24:24 UTC+2, Timothy Snyder a écrit :

This page goes into a little more details than the VGSL spec page in the Tesseract repo: https://github.com/mldbai/tensorflow-models/blob/master/street/g3doc/vgslspecs.md

Not specific to Tesseract but this guy's articles have good info on lower-level neural net mechanics in relation to OCR.

On Fri, Sep 6, 2019 at 9:12 AM Purushotham Rao Eravalli <purus...@sukshi.com> wrote:

It will be great if you provide any source where we can get detailed information about the architecture used for tesseract and it's loss functions or so.

Thanks

On Fri, Sep 6, 2019, 6:39 PM Timothy Snyder <tc...@zips.uakron.edu> wrote:

Do you want to learn more about neural networks or specifically, a "summarizing LSTM" in a neural network?

On Fri, Sep 6, 2019 at 5:05 AM Youcef <youce...@gmail.com> wrote:

Hi,

In that page https://github.com/tesseract-ocr/tesseract/wiki/VGSLSpecs from officiel github repo, it talks about "summarizing LSTM" and how we can manage to recognize image of both variable height and variable width . Is there a paper, a blog or anything else that provide more explicit information regarding that, that provide explicit architecture examples? I didn't succeed in google it, maybe I'm not using the right keywords. 

Thanks for any help.

Regards

Youcef

--
You received this message because you are subscribed to the Google Groups "tesseract-ocr" group.

To unsubscribe from this group and stop receiving emails from it, send an email to tesser...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/0a940ac0-9000-46c6-9f93-73c78c29068e%40googlegroups.com.

--
You received this message because you are subscribed to the Google Groups "tesseract-ocr" group.

To unsubscribe from this group and stop receiving emails from it, send an email to tesser...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/CABtjQ9KCH-X%3DNLmjdaPMsX9jefMDPpyFD-Sj-XSvWoQURqUvoA%40mail.gmail.com.

--
You received this message because you are subscribed to the Google Groups "tesseract-ocr" group.

To unsubscribe from this group and stop receiving emails from it, send an email to tesser...@googlegroups.com.