> [...]
> > +#define OFFSET(x) offsetof(SRCNNContext, x)
> > +#define FLAGS AV_OPT_FLAG_FILTERING_PARAM | AV_OPT_FLAG_VIDEO_PARAM
> > +static const AVOption srcnn_options[] = {
> > +{ "config_file", "path to configuration file with network
> parameters", OFFSET(config_file_path), AV_OPT_TYPE_ST
2018-05-07 17:41 GMT+03:00 Pedro Arthur :
> 2018-05-07 0:30 GMT-03:00 Steven Liu :
> > Hi Sergey,
> >
> > How should i test this filter?
> > I tested it some days ago, the picture get worse from 2nd frame.
> > input resolution 640x480 to 1280x720;
> >
> > ffmpeg -i
2018-05-24 22:52 GMT+03:00 James Almer :
> On 5/24/2018 4:24 PM, Sergey Lavrushkin wrote:
> > Hello,
> >
> > This patch introduces DNN inference interface and simple native backend.
> > For now implemented backend supports only convolutions with relu
> activation
>
2018-05-29 4:08 GMT+03:00 Pedro Arthur :
> 2018-05-28 19:52 GMT-03:00 Sergey Lavrushkin :
> > 2018-05-28 9:32 GMT+03:00 Guo, Yejun :
> >
> >> looks that no tensorflow dependency is introduced, a new model format is
> >> created together with some CPU implemen
2018-06-01 6:09 GMT+03:00 Guo, Yejun :
> Did you try to build ffmpeg with TENSORFLOW_BACKEND enabled, and run it
> without TF library? This case is possible when an end user install
> pre-built package on a machine without TF library.
>
> In function init, the logic is to fall back to cpu path (D
2018-06-02 19:45 GMT+03:00 James Almer :
> On 5/31/2018 12:01 PM, Sergey Lavrushkin wrote:
> > diff --git a/Changelog b/Changelog
> > index df2024fb59..a667fd045d 100644
> > --- a/Changelog
> > +++ b/Changelog
> > @@ -11,6 +11,7 @@ version :
> > - support
2018-06-03 19:57 GMT+03:00 Pedro Arthur :
> 2018-05-31 12:01 GMT-03:00 Sergey Lavrushkin :
> > Hello,
> >
> > This patch introduces TensorFlow backend for DNN inference module.
> > This backend uses TensorFlow binary models and requires from model
> > to have th
>
> My concern is when we add more models, currently we have to store 2
> models, one for the "native" implementation and one for the TF
> backend.
> There is also the case were one wants to update the weights for a
> model, it will be necessary to update both the native and TF data.
> Having dupli
2018-06-05 17:20 GMT+03:00 James Almer :
> On 6/3/2018 3:02 PM, Sergey Lavrushkin wrote:
> > diff --git a/libavfilter/vf_srcnn.c b/libavfilter/vf_srcnn.c
> > index d6efe9b478..5c5e26b33a 100644
> > --- a/libavfilter/vf_srcnn.c
> > +++ b/libavfilter/vf_srcnn.c
> >
2018-06-06 17:22 GMT+03:00 Pedro Arthur :
> Hi,
>
> 2018-06-05 20:23 GMT-03:00 Sergey Lavrushkin :
> > Here is the patch, that fixes described issues.
> When I try to run (video input), when tf is not enabled in configure it
> crashes.
>
>
> $ffmpeg -i in.mp4 -v
2018-07-28 4:31 GMT+03:00 Michael Niedermayer :
> On Fri, Jul 27, 2018 at 08:06:15PM +0300, Sergey Lavrushkin wrote:
> > Hello,
> >
> > The first patch provides on the fly generation of default DNN models,
> > that eliminates data duplication for model weights. Al
2018-07-30 2:01 GMT+03:00 Michael Niedermayer :
> On Sat, Jul 28, 2018 at 01:00:53PM +0300, Sergey Lavrushkin wrote:
> > 2018-07-28 4:31 GMT+03:00 Michael Niedermayer :
> >
> > > On Fri, Jul 27, 2018 at 08:06:15PM +0300, Sergey Lavrushkin wrote:
> > > > He
---
libavfilter/dnn_espcn.h | 3 ++-
libavfilter/dnn_srcnn.h | 3 ++-
2 files changed, 4 insertions(+), 2 deletions(-)
diff --git a/libavfilter/dnn_espcn.h b/libavfilter/dnn_espcn.h
index 9344aa90fe..e0013fe1dd 100644
--- a/libavfilter/dnn_espcn.h
+++ b/libavfilter/dnn_espcn.h
@@ -20,7 +20,8 @@
---
libavfilter/dnn_backend_tf.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/libavfilter/dnn_backend_tf.c b/libavfilter/dnn_backend_tf.c
index 7a4ad72d27..662a2a3c6e 100644
--- a/libavfilter/dnn_backend_tf.c
+++ b/libavfilter/dnn_backend_tf.c
@@ -570,7 +570,9 @@ void ff_
This patch removes conversions, declared inside the sr filter, and uses
libswscale inside
the filter to perform them for only Y channel of input. The sr filter still has
uint
formats as input, as it does not use chroma channels in models and these
channels are
upscaled using libswscale, float fo
---
libavfilter/dnn_backend_tf.c | 64 +++-
libavfilter/dnn_espcn.h | 37 -
libavfilter/dnn_srcnn.h | 35
3 files changed, 63 insertions(+), 73 deletions(-)
diff --git a/libavfilter/dnn_backend_tf.c
Hello,
These patches address several raised concerns regarding sr filter
and DNN module. I included three patches, that I've already sent,
but they still have not been reviewed properly.
libavfilter: Adds on the fly generation of default DNN models for
tensorflow backend instead of storing
---
libavfilter/dnn_backend_native.c | 84 +++---
libavfilter/dnn_backend_native.h | 8 +--
libavfilter/dnn_backend_tf.c | 108 +++
libavfilter/dnn_backend_tf.h | 8 +--
libavfilter/dnn_espcn.h | 6 +--
libavfilter
This patch adds two floating-point gray formats to use them in sr filter for
conversion with libswscale. I added conversion from uint gray to float and
backwards in swscale_unscaled.c, that is enough for sr filter. But for
proper format addition, should I add anything else?
---
libavutil/pixdesc
2018-08-03 16:07 GMT+03:00 Michael Niedermayer :
> On Thu, Aug 02, 2018 at 09:52:45PM +0300, Sergey Lavrushkin wrote:
> > This patch adds two floating-point gray formats to use them in sr filter
> for
> > conversion with libswscale. I added conversion from uint gray to float
&
2018-08-04 0:11 GMT+03:00 Michael Niedermayer :
> On Fri, Aug 03, 2018 at 10:33:00PM +0300, Sergey Lavrushkin wrote:
> > 2018-08-03 16:07 GMT+03:00 Michael Niedermayer :
> >
> > > On Thu, Aug 02, 2018 at 09:52:45PM +0300, Sergey Lavrushkin wrote:
> > > > Thi
Updated patch.
2018-08-06 17:55 GMT+03:00 Pedro Arthur :
> 2018-08-02 15:52 GMT-03:00 Sergey Lavrushkin :
> > ---
> > libavfilter/dnn_backend_native.c | 84 +++---
> > libavfilter/dnn_backend_native.h | 8 +--
> > libavfilter
Made variables static.
2018-08-06 21:19 GMT+03:00 Pedro Arthur :
> 2018-08-02 15:52 GMT-03:00 Sergey Lavrushkin :
> > ---
> > libavfilter/dnn_backend_tf.c | 64 ++
> +-
> > libavfilter/dnn_espcn.h | 37 --
I split patch to one for libavutil and another for libswscale,
also added LUT for unscaled conversion, added
conversions for scaling and updated fate tests.
From 8bcc10b49c41612b4d6549e64d90acf3f0b3fc6a Mon Sep 17 00:00:00 2001
From: Sergey Lavrushkin
Date: Fri, 3 Aug 2018 18:02:49 +0300
Subject
Updated patch.
From 11186187d0b5a4725415a91947f38d5e166e024c Mon Sep 17 00:00:00 2001
From: Sergey Lavrushkin
Date: Tue, 31 Jul 2018 18:40:24 +0300
Subject: [PATCH 6/9] libavfilter/dnn_backend_tf.c: Fixes ff_dnn_free_model_tf.
---
libavfilter/dnn_backend_tf.c | 4 +++-
1 file changed, 3
Updated patch.
2018-08-02 21:52 GMT+03:00 Sergey Lavrushkin :
> ---
> libavfilter/dnn_espcn.h | 3 ++-
> libavfilter/dnn_srcnn.h | 3 ++-
> 2 files changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/libavfilter/dnn_espcn.h b/libavfilter/dnn_espcn.h
> index 9344aa90
From f076c4be5455331958b928fcea6b3dd8da287527 Mon Sep 17 00:00:00 2001
From: Sergey Lavrushkin
Date: Fri, 3 Aug 2018 17:24:00 +0300
Subject: [PATCH 9/9] doc/filters.texi: Adds documentation for sr filter.
---
doc/filters.texi | 60
1
Updated patch.
2018-08-02 21:52 GMT+03:00 Sergey Lavrushkin :
> This patch removes conversions, declared inside the sr filter, and uses
> libswscale inside
> the filter to perform them for only Y channel of input. The sr filter
> still has uint
> formats as input, as it does
Here are updated patches with fixes. I updated conversion functions, so
they should
properly work with format for different endianness.
2018-08-08 1:47 GMT+03:00 Michael Niedermayer :
> On Tue, Aug 07, 2018 at 12:17:58AM +0300, Sergey Lavrushkin wrote:
> > I split patch to one for liba
2018-08-07 13:14 GMT+03:00 Moritz Barsnick :
> On Tue, Aug 07, 2018 at 00:24:29 +0300, Sergey Lavrushkin wrote:
> > +@table @option
> > +@item model
> > +Specify what super-resolution model to use. This option accepts the
> following values:
>^ nit: wh
Sorry, I accidentally sent previous patch, here is updated version.
From 99afeefe4add5b932140388f48ec4111734aa593 Mon Sep 17 00:00:00 2001
From: Sergey Lavrushkin
Date: Fri, 3 Aug 2018 17:24:00 +0300
Subject: [PATCH 9/9] doc/filters.texi: Adds documentation for sr filter.
---
doc/filters.texi
2018-08-10 20:24 GMT+03:00 Michael Niedermayer :
> On Thu, Aug 09, 2018 at 08:15:16PM +0300, Sergey Lavrushkin wrote:
> > Here are updated patches with fixes. I updated conversion functions, so
> > they should
> > properly work with format for different endianness.
>
2018-08-12 0:45 GMT+03:00 Michael Niedermayer :
> On Sat, Aug 11, 2018 at 05:52:32PM +0300, Sergey Lavrushkin wrote:
> > 2018-08-10 20:24 GMT+03:00 Michael Niedermayer :
> >
> > > On Thu, Aug 09, 2018 at 08:15:16PM +0300, Sergey Lavrushkin wrote:
> > > > He
>
> Just use av_clipf instead of FFMIN/FFMAX.
Changed FFMIN/FFMAX to av_clip_uint16 and av_clip_uint8.
From 210e497d76328947fdf424b169728fa728cc18f2 Mon Sep 17 00:00:00 2001
From: Sergey Lavrushkin
Date: Fri, 3 Aug 2018 18:06:50 +0300
Subject: [PATCH 5/9] libswscale: Adds conversions f
2018-08-15 1:49 GMT+03:00 Marton Balint :
>
> On Tue, 14 Aug 2018, Pedro Arthur wrote:
>
> 2018-08-14 15:45 GMT-03:00 Rostislav Pehlivanov :
>>
>>> On Thu, 2 Aug 2018 at 20:00, Sergey Lavrushkin
>>> wrote:
>>>
>>> This patch remove
Resending patch with documentation for sr filter.
---
doc/filters.texi | 60
1 file changed, 60 insertions(+)
diff --git a/doc/filters.texi b/doc/filters.texi
index 267bd04a43..b2a74cb1ce 100644
--- a/doc/filters.texi
+++ b/doc/filters.tex
2018-08-15 19:59 GMT+03:00 Gyan Doshi :
>
>
> On 15-08-2018 10:05 PM, Sergey Lavrushkin wrote:
>
>> Resending patch with documentation for sr filter.
>>
>
> LGTM. Will apply with some small changes.
>
> I've merged the docs entry in the 2nd part, so remove
пт, 17 авг. 2018 г., 6:47 James Almer :
> On 8/14/2018 1:23 PM, Michael Niedermayer wrote:
> > On Mon, Aug 13, 2018 at 04:58:42PM +0300, Sergey Lavrushkin wrote:
> >>>
> >>> Just use av_clipf instead of FFMIN/FFMAX.
> >>
> >>
> >
2018-08-17 17:46 GMT+03:00 Pedro Arthur :
> Hi,
>
> You did not provided any pre trained model files, so anyone trying to
> test it has to perform the whole training!
> I'm attaching the models I generated, if anyone is interested in testing
> it.
>
> When applying the filter with tf backend there
2018-08-17 23:28 GMT+03:00 Michael Niedermayer :
> On Fri, Aug 17, 2018 at 12:46:52AM -0300, James Almer wrote:
> > On 8/14/2018 1:23 PM, Michael Niedermayer wrote:
> > > On Mon, Aug 13, 2018 at 04:58:42PM +0300, Sergey Lavrushkin wrote:
> > >>>
> > >&
2018-08-18 23:20 GMT+03:00 Michael Niedermayer :
> On Sat, Aug 18, 2018 at 02:10:21PM +0300, Sergey Lavrushkin wrote:
> > 2018-08-17 23:28 GMT+03:00 Michael Niedermayer :
> >
> > > On Fri, Aug 17, 2018 at 12:46:52AM -0300, James Almer wrote:
> > > > On 8/14/201
41 matches
Mail list logo
