[FFmpeg-devel,2/2] doc/filters: update how to generate native model for sr filter

Submitted by Guo, Yejun on May 28, 2019, 8:01 a.m.

Details

Message ID 1559030485-894-1-git-send-email-yejun.guo@intel.com
State New
Headers show

Commit Message

Guo, Yejun May 28, 2019, 8:01 a.m.
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
---
 doc/filters.texi | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

Comments

Gyan May 28, 2019, 8:14 a.m.
On 28-05-2019 01:31 PM, Guo, Yejun wrote:
> Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
> ---
>   doc/filters.texi | 5 +++--
>   1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/doc/filters.texi b/doc/filters.texi
> index 4fdcfe9..75d2a38 100644
> --- a/doc/filters.texi
> +++ b/doc/filters.texi
> @@ -16538,9 +16538,10 @@ Efficient Sub-Pixel Convolutional Neural Network model (ESPCN).
>   See @url{https://arxiv.org/abs/1609.05158}.
>   @end itemize
>   
> -Training scripts as well as scripts for model generation can be found at
> +Training scripts as well as scripts for model file (.pb) saving can be found at
What's wrong with generation'?

>   @url{https://github.com/XueweiMeng/sr/tree/sr_dnn_native}. Original repository
> -is at @url{https://github.com/HighVoltageRocknRoll/sr.git}.
> +is at @url{https://github.com/HighVoltageRocknRoll/sr.git}. Once get the TensorFlow
> +model file (.pb), the native model file (.model) can be generated via libavfilter/dnn/python/convert.py
Change to

"Native model files (.model) can be generated from TensorFlow model 
files (.pb) by using libavfilter/dnn/python/convert.py "

and shift to a new line.

Gyan
Guo, Yejun May 28, 2019, 8:32 a.m.
> -----Original Message-----

> From: ffmpeg-devel [mailto:ffmpeg-devel-bounces@ffmpeg.org] On Behalf Of

> Gyan

> Sent: Tuesday, May 28, 2019 4:14 PM

> To: ffmpeg-devel@ffmpeg.org

> Subject: Re: [FFmpeg-devel] [PATCH 2/2] doc/filters: update how to generate

> native model for sr filter

> 

> 

> 

> On 28-05-2019 01:31 PM, Guo, Yejun wrote:

> > Signed-off-by: Guo, Yejun <yejun.guo@intel.com>

> > ---

> >   doc/filters.texi | 5 +++--

> >   1 file changed, 3 insertions(+), 2 deletions(-)

> >

> > diff --git a/doc/filters.texi b/doc/filters.texi

> > index 4fdcfe9..75d2a38 100644

> > --- a/doc/filters.texi

> > +++ b/doc/filters.texi

> > @@ -16538,9 +16538,10 @@ Efficient Sub-Pixel Convolutional Neural

> Network model (ESPCN).

> >   See @url{https://arxiv.org/abs/1609.05158}.

> >   @end itemize

> >

> > -Training scripts as well as scripts for model generation can be found at

> > +Training scripts as well as scripts for model file (.pb) saving can be found at

> What's wrong with generation'?


Currently, the script generates both model file formats (.pb and .model). The TensorFlow model .pb
file will not change again, while the native .model file format will be changed timely especially in current
early stage.  It is not good to generate .model file here (see more detail in my patch 1 commit log). 

So, a new script (my patch 1) is added to generate the .model file. The above script is no longer needed to
generate .model file. 

For me, 'generation' means the dump of .model file, while 'saving' means the dump of .pb file.

> 

> >   @url{https://github.com/XueweiMeng/sr/tree/sr_dnn_native}. Original

> repository

> > -is at @url{https://github.com/HighVoltageRocknRoll/sr.git}.

> > +is at @url{https://github.com/HighVoltageRocknRoll/sr.git}. Once get the

> TensorFlow

> > +model file (.pb), the native model file (.model) can be generated via

> libavfilter/dnn/python/convert.py

> Change to

> 

> "Native model files (.model) can be generated from TensorFlow model

> files (.pb) by using libavfilter/dnn/python/convert.py "

> 

> and shift to a new line.


thanks, will do it.

> 

> Gyan

> _______________________________________________

> ffmpeg-devel mailing list

> ffmpeg-devel@ffmpeg.org

> https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

> 

> To unsubscribe, visit link above, or email

> ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

Patch hide | download patch | download mbox

diff --git a/doc/filters.texi b/doc/filters.texi
index 4fdcfe9..75d2a38 100644
--- a/doc/filters.texi
+++ b/doc/filters.texi
@@ -16538,9 +16538,10 @@  Efficient Sub-Pixel Convolutional Neural Network model (ESPCN).
 See @url{https://arxiv.org/abs/1609.05158}.
 @end itemize
 
-Training scripts as well as scripts for model generation can be found at
+Training scripts as well as scripts for model file (.pb) saving can be found at
 @url{https://github.com/XueweiMeng/sr/tree/sr_dnn_native}. Original repository
-is at @url{https://github.com/HighVoltageRocknRoll/sr.git}.
+is at @url{https://github.com/HighVoltageRocknRoll/sr.git}. Once get the TensorFlow
+model file (.pb), the native model file (.model) can be generated via libavfilter/dnn/python/convert.py
 
 The filter accepts the following options: