[FFmpeg-devel] [PATCH V2 4/4] avfilter/vf_dnn_processing: add a generic filter for image proccessing with dnn networks

Paul B Mahol onemda at gmail.com
Mon Oct 28 13:51:27 EET 2019


On 10/28/19, Guo, Yejun <yejun.guo at intel.com> wrote:
>
>
>> -----Original Message-----
>> From: Paul B Mahol [mailto:onemda at gmail.com]
>> Sent: Monday, October 28, 2019 4:00 PM
>> To: FFmpeg development discussions and patches <ffmpeg-devel at ffmpeg.org>
>> Cc: Guo, Yejun <yejun.guo at intel.com>
>> Subject: Re: [FFmpeg-devel] [PATCH V2 4/4] avfilter/vf_dnn_processing: add
>> a
>> generic filter for image proccessing with dnn networks
>>
>> On 10/21/19, Guo, Yejun <yejun.guo at intel.com> wrote:
>> > This filter accepts all the dnn networks which do image processing.
>> > Currently, frame with formats rgb24 and bgr24 are supported. Other
>> > formats such as gray and YUV will be supported next. The dnn network
>> > can accept data in float32 or uint8 format. And the dnn network can
>> > change frame size.
>> >
>> > Let's take an example with the following python script. This script
>> > halves the value of the first channel of the pixel.
>> > import tensorflow as tf
>> > import numpy as np
>> > import scipy.misc
>> > in_img = scipy.misc.imread('in.bmp')
>> > in_img = in_img.astype(np.float32)/255.0
>> > in_data = in_img[np.newaxis, :]
>> > filter_data = np.array([0.5, 0, 0, 0, 1., 0, 0, 0,
>> > 1.]).reshape(1,1,3,3).astype(np.float32)
>> > filter = tf.Variable(filter_data)
>> > x = tf.placeholder(tf.float32, shape=[1, None, None, 3], name='dnn_in')
>> > y = tf.nn.conv2d(x, filter, strides=[1, 1, 1, 1], padding='VALID',
>> > name='dnn_out')
>> > sess=tf.Session()
>> > sess.run(tf.global_variables_initializer())
>> > output = sess.run(y, feed_dict={x: in_data})
>> > graph_def = tf.graph_util.convert_variables_to_constants(sess,
>> > sess.graph_def, ['dnn_out'])
>> > tf.train.write_graph(graph_def, '.', 'halve_first_channel.pb',
>> > as_text=False)
>> > output = output * 255.0
>> > output = output.astype(np.uint8)
>> > scipy.misc.imsave("out.bmp", np.squeeze(output))
>>
>> So this one executes python code without ever returning or using AVFrame*
>> ?
>> This is extremely limited usage.
>
> the purpose of this script is to demo how to setup and execute dnn models
> with python+tensorflow.
> The only relationship with ffmpeg is that the script prepares the model file
> halve_first_channel.pb.
>
> The next description shows how ffmpeg can execute the model in a filter.
>
> I'll try to update the commit log to avoid misleading words, thanks.
>
>>
>> >
>> > - generate halve_first_channel.pb with the above script
>> > - generate halve_first_channel.model with tools/python/convert.py
>> > - try with following commands
>> >   ./ffmpeg -i input.jpg -vf
>> >
>> dnn_processing=model=halve_first_channel.model:input=dnn_in:output=dnn_
>> out:fmt=rgb24:dnn_backend=native
>> > -y out.native.png
>> >   ./ffmpeg -i input.jpg -vf
>> >
>> dnn_processing=model=halve_first_channel.pb:input=dnn_in:output=dnn_out:f
>> mt=rgb24:dnn_backend=tensorflow
>> > -y out.tf.png
>> >
>> > Signed-off-by: Guo, Yejun <yejun.guo at intel.com>
>> > ---
>> >  configure                       |   1 +
>> >  doc/filters.texi                |  44 ++++++
>> >  libavfilter/Makefile            |   1 +
>> >  libavfilter/allfilters.c        |   1 +
>> >  libavfilter/vf_dnn_processing.c | 333
>> > ++++++++++++++++++++++++++++++++++++++++
>> >  5 files changed, 380 insertions(+)
>> >  create mode 100644 libavfilter/vf_dnn_processing.c
>> >
>> > diff --git a/configure b/configure
>> > index 8413826..bf2bac9 100755
>> > --- a/configure
>> > +++ b/configure
>> > @@ -3460,6 +3460,7 @@ derain_filter_select="dnn"
>> >  deshake_filter_select="pixelutils"
>> >  deshake_opencl_filter_deps="opencl"
>> >  dilation_opencl_filter_deps="opencl"
>> > +dnn_processing_filter_select="dnn"
>> >  drawtext_filter_deps="libfreetype"
>> >  drawtext_filter_suggest="libfontconfig libfribidi"
>> >  elbg_filter_deps="avcodec"
>> > diff --git a/doc/filters.texi b/doc/filters.texi
>> > index bdc4136..c11a616 100644
>> > --- a/doc/filters.texi
>> > +++ b/doc/filters.texi
>> > @@ -8928,6 +8928,50 @@ ffmpeg -i INPUT -f lavfi -i
>> > nullsrc=hd720,geq='r=128+80*(sin(sqrt((X-W/2)*(X-W/2
>> >  @end example
>> >  @end itemize
>> >
>> > + at section dnn_processing
>> > +
>> > +Do image processing with deep neural networks. Currently only AVFrame
>> with
>> > RGB24
>> > +and BGR24 are supported, more formats will be added later.
>> > +
>> > +The filter accepts the following options:
>> > +
>> > + at table @option
>> > + at item dnn_backend
>> > +Specify which DNN backend to use for model loading and execution. This
>> > option accepts
>> > +the following values:
>> > +
>> > + at table @samp
>> > + at item native
>> > +Native implementation of DNN loading and execution.
>> > +
>> > + at item tensorflow
>> > +TensorFlow backend. To enable this backend you
>> > +need to install the TensorFlow for C library (see
>> > + at url{https://www.tensorflow.org/install/install_c}) and configure
>> > FFmpeg
>> > with
>> > + at code{--enable-libtensorflow}
>> > + at end table
>> > +
>> > +Default value is @samp{native}.
>> > +
>> > + at item model
>> > +Set path to model file specifying network architecture and its
>> > parameters.
>> > +Note that different backends use different file formats. TensorFlow
>> > and
>> > native
>> > +backend can load files for only its format.
>> > +
>> > +Native model file (.model) can be generated from TensorFlow model file
>> > (.pb) by using tools/python/convert.py
>> > +
>> > + at item input
>> > +Set the input name of the dnn network.
>> > +
>> > + at item output
>> > +Set the output name of the dnn network.
>> > +
>> > + at item fmt
>> > +Set the pixel format for the Frame. Allowed values are
>> > @code{AV_PIX_FMT_RGB24}, and @code{AV_PIX_FMT_BGR24}.
>> > +Default value is @code{AV_PIX_FMT_RGB24}.
>> > +
>> > + at end table
>> > +
>> >  @section drawbox
>> >
>> >  Draw a colored box on the input image.
>> > diff --git a/libavfilter/Makefile b/libavfilter/Makefile
>> > index 63d2fba..47a485a 100644
>> > --- a/libavfilter/Makefile
>> > +++ b/libavfilter/Makefile
>> > @@ -224,6 +224,7 @@ OBJS-$(CONFIG_DILATION_OPENCL_FILTER)
>> +=
>> > vf_neighbor_opencl.o opencl.o \
>> >
>> opencl/neighbor.o
>> >  OBJS-$(CONFIG_DISPLACE_FILTER)               += vf_displace.o
>> framesync.o
>> >  OBJS-$(CONFIG_DOUBLEWEAVE_FILTER)            += vf_weave.o
>> > +OBJS-$(CONFIG_DNN_PROCESSING_FILTER)         +=
>> vf_dnn_processing.o
>> >  OBJS-$(CONFIG_DRAWBOX_FILTER)                += vf_drawbox.o
>> >  OBJS-$(CONFIG_DRAWGRAPH_FILTER)              += f_drawgraph.o
>> >  OBJS-$(CONFIG_DRAWGRID_FILTER)               += vf_drawbox.o
>> > diff --git a/libavfilter/allfilters.c b/libavfilter/allfilters.c
>> > index e4186f9..485409f 100644
>> > --- a/libavfilter/allfilters.c
>> > +++ b/libavfilter/allfilters.c
>> > @@ -209,6 +209,7 @@ extern AVFilter ff_vf_detelecine;
>> >  extern AVFilter ff_vf_dilation;
>> >  extern AVFilter ff_vf_dilation_opencl;
>> >  extern AVFilter ff_vf_displace;
>> > +extern AVFilter ff_vf_dnn_processing;
>> >  extern AVFilter ff_vf_doubleweave;
>> >  extern AVFilter ff_vf_drawbox;
>> >  extern AVFilter ff_vf_drawgraph;
>> > diff --git a/libavfilter/vf_dnn_processing.c
>> > b/libavfilter/vf_dnn_processing.c
>> > new file mode 100644
>> > index 0000000..de89af4
>> > --- /dev/null
>> > +++ b/libavfilter/vf_dnn_processing.c
>> > @@ -0,0 +1,333 @@
>> > +/*
>> > + * Copyright (c) 2019 Guo Yejun
>> > + *
>> > + * This file is part of FFmpeg.
>> > + *
>> > + * FFmpeg is free software; you can redistribute it and/or
>> > + * modify it under the terms of the GNU Lesser General Public
>> > + * License as published by the Free Software Foundation; either
>> > + * version 2.1 of the License, or (at your option) any later version.
>> > + *
>> > + * FFmpeg is distributed in the hope that it will be useful,
>> > + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>> > + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
>> GNU
>> > + * Lesser General Public License for more details.
>> > + *
>> > + * You should have received a copy of the GNU Lesser General Public
>> > + * License along with FFmpeg; if not, write to the Free Software
>> > + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
>> > 02110-1301
>> > USA
>> > + */
>> > +
>> > +/**
>> > + * @file
>> > + * implementing a generic image processing filter using deep learning
>> > networks.
>> > + */
>> > +
>> > +#include "libavformat/avio.h"
>> > +#include "libavutil/opt.h"
>> > +#include "libavutil/pixdesc.h"
>> > +#include "libavutil/avassert.h"
>> > +#include "avfilter.h"
>> > +#include "dnn_interface.h"
>> > +#include "formats.h"
>> > +#include "internal.h"
>> > +
>> > +typedef struct DnnProcessingContext {
>> > +    const AVClass *class;
>> > +
>> > +    char *model_filename;
>> > +    DNNBackendType backend_type;
>> > +    enum AVPixelFormat fmt;
>>
>> This should be int.
>
> could you please help to explain a bit more why 'enum AVPixelFormat' should
> be int.
>
> I searched 'AV_OPT_TYPE_PIXEL_FMT' in vf_* files and found 'enum
> AVPixelFormat' is used in
> vf_mergeplanes.c, vf_program_opencl.c and vf_tonemap_opencl.c.

That is error, I gonna fix mergeplanes ASAP. Sometimes enum may use
completely another type, Michael may know more.
Besides options assumes ints and not enums.

>
>>
>> > +    char *model_inputname;
>> > +    char *model_outputname;
>> > +
>> > +    DNNModule *dnn_module;
>> > +    DNNModel *model;
>> > +
>> > +    // input & output of the model at execution time
>> > +    DNNData input;
>> > +    DNNData output;
>> > +} DnnProcessingContext;
>> > +
>> > +#define OFFSET(x) offsetof(DnnProcessingContext, x)
>> > +#define FLAGS AV_OPT_FLAG_FILTERING_PARAM |
>> AV_OPT_FLAG_VIDEO_PARAM
>> > +static const AVOption dnn_processing_options[] = {
>> > +    { "dnn_backend", "DNN backend",
>> OFFSET(backend_type),
>> >  AV_OPT_TYPE_INT,       { .i64 = 0 },    0, 1, FLAGS, "backend" },
>> > +    { "native",      "native backend flag",        0,
>> >  AV_OPT_TYPE_CONST,     { .i64 = 0 },    0, 0, FLAGS, "backend" },
>> > +#if (CONFIG_LIBTENSORFLOW == 1)
>> > +    { "tensorflow",  "tensorflow backend flag",    0,
>> >  AV_OPT_TYPE_CONST,     { .i64 = 1 },    0, 0, FLAGS, "backend" },
>> > +#endif
>> > +    { "model",       "path to model file",
>> OFFSET(model_filename),
>> >  AV_OPT_TYPE_STRING,    { .str = NULL }, 0, 0, FLAGS },
>> > +    { "input",       "input name of the model",
>> OFFSET(model_inputname),
>> >  AV_OPT_TYPE_STRING,    { .str = NULL }, 0, 0, FLAGS },
>> > +    { "output",      "output name of the model",
>> > OFFSET(model_outputname), AV_OPT_TYPE_STRING,    { .str = NULL }, 0, 0,
>> > FLAGS },
>> > +    { "fmt",         "AVPixelFormat of the frame", OFFSET(fmt),
>> >  AV_OPT_TYPE_PIXEL_FMT, { .i64=AV_PIX_FMT_RGB24 },
>> AV_PIX_FMT_NONE,
>> > AV_PIX_FMT_NB - 1, FLAGS },
>> > +    { NULL }
>> > +};
>> > +
>> > +AVFILTER_DEFINE_CLASS(dnn_processing);
>> > +
>> > +static av_cold int init(AVFilterContext *context)
>> > +{
>> > +    DnnProcessingContext *ctx = context->priv;
>> > +    int supported = 0;
>> > +    // as the first step, only rgb24 and bgr24 are supported
>> > +    const enum AVPixelFormat supported_pixel_fmts[] = {
>> > +        AV_PIX_FMT_RGB24,
>> > +        AV_PIX_FMT_BGR24,
>> > +    };
>> > +    for (int i = 0; i < sizeof(supported_pixel_fmts) / sizeof(enum
>> > AVPixelFormat); ++i) {
>> > +        if (supported_pixel_fmts[i] == ctx->fmt) {
>> > +            supported = 1;
>> > +            break;
>> > +        }
>> > +    }
>> > +    if (!supported) {
>> > +        av_log(context, AV_LOG_ERROR, "pixel fmt %s not supported
>> yet\n",
>> > +
>> av_get_pix_fmt_name(ctx->fmt));
>> > +        return AVERROR(AVERROR_INVALIDDATA);
>> > +    }
>> > +
>> > +    if (!ctx->model_filename) {
>> > +        av_log(ctx, AV_LOG_ERROR, "model file for network is not
>> > specified\n");
>> > +        return AVERROR(EINVAL);
>> > +    }
>> > +    if (!ctx->model_inputname) {
>> > +        av_log(ctx, AV_LOG_ERROR, "intput name of the model network is
>> not
>>
>> Typo
>
> thanks, will fix.
>
>>
>> > specified\n");
>> > +        return AVERROR(EINVAL);
>> > +    }
>> > +    if (!ctx->model_outputname) {
>> > +        av_log(ctx, AV_LOG_ERROR, "output name of the model network
>> is not
>> > specified\n");
>> > +        return AVERROR(EINVAL);
>> > +    }
>> > +
>> > +    ctx->dnn_module = ff_get_dnn_module(ctx->backend_type);
>> > +    if (!ctx->dnn_module) {
>> > +        av_log(ctx, AV_LOG_ERROR, "could not create DNN module for
>> > requested backend\n");
>> > +        return AVERROR(ENOMEM);
>> > +    }
>> > +    if (!ctx->dnn_module->load_model) {
>> > +        av_log(ctx, AV_LOG_ERROR, "load_model for network is not
>> > specified\n");
>> > +        return AVERROR(EINVAL);
>> > +    }
>> > +
>> > +    ctx->model = (ctx->dnn_module->load_model)(ctx->model_filename);
>> > +    if (!ctx->model) {
>> > +        av_log(ctx, AV_LOG_ERROR, "could not load DNN model\n");
>> > +        return AVERROR(EINVAL);
>> > +    }
>> > +
>> > +    return 0;
>> > +}
>> > +
>> > +static int query_formats(AVFilterContext *context)
>> > +{
>> > +    AVFilterFormats *formats;
>> > +    DnnProcessingContext *ctx = context->priv;
>> > +    enum AVPixelFormat pixel_fmts[2];
>> > +    pixel_fmts[0] = ctx->fmt;
>> > +    pixel_fmts[1] = AV_PIX_FMT_NONE;
>> > +
>> > +    formats = ff_make_format_list(pixel_fmts);
>> > +    return ff_set_common_formats(context, formats);
>> > +}
>> > +
>> > +static int config_input(AVFilterLink *inlink)
>> > +{
>> > +    AVFilterContext *context     = inlink->dst;
>> > +    DnnProcessingContext *ctx = context->priv;
>> > +    DNNReturnType result;
>> > +    DNNData dnn_data;
>> > +
>> > +    result = ctx->model->get_input(ctx->model->model, &dnn_data,
>> > ctx->model_inputname);
>> > +    if (result != DNN_SUCCESS) {
>> > +        av_log(ctx, AV_LOG_ERROR, "could not get input from the
>> model\n");
>> > +        return AVERROR(EIO);
>> > +    }
>> > +
>> > +    // the design is to add explicit scale filter before this filter
>> > +    if (dnn_data.height != -1 && dnn_data.height != inlink->h) {
>> > +        av_log(ctx, AV_LOG_ERROR, "the model requires frame height %d
>> but
>> > got %d\n",
>> > +                                   dnn_data.height, inlink->h);
>> > +        return AVERROR(EIO);
>> > +    }
>> > +    if (dnn_data.width != -1 && dnn_data.width != inlink->w) {
>> > +        av_log(ctx, AV_LOG_ERROR, "the model requires frame width %d
>> but
>> > got %d\n",
>> > +                                   dnn_data.width, inlink->w);
>> > +        return AVERROR(EIO);
>> > +    }
>> > +
>> > +    if (dnn_data.channels != 3) {
>> > +        av_log(ctx, AV_LOG_ERROR, "the model requires input
>> channels %d\n",
>> > +                                   dnn_data.channels);
>> > +        return AVERROR(EIO);
>> > +    }
>> > +    if (dnn_data.dt != DNN_FLOAT && dnn_data.dt != DNN_UINT8) {
>> > +        av_log(ctx, AV_LOG_ERROR, "only support dnn models with input
>> data
>> > type as float32 and uint8.\n");
>> > +        return AVERROR(EIO);
>> > +    }
>> > +
>> > +    ctx->input.width    = inlink->w;
>> > +    ctx->input.height   = inlink->h;
>> > +    ctx->input.channels = dnn_data.channels;
>> > +    ctx->input.dt = dnn_data.dt;
>> > +
>> > +    result = (ctx->model->set_input_output)(ctx->model->model,
>> > +                                        &ctx->input,
>> ctx->model_inputname,
>> > +                                        (const char
>> > **)&ctx->model_outputname, 1);
>> > +    if (result != DNN_SUCCESS) {
>> > +        av_log(ctx, AV_LOG_ERROR, "could not set input and output for
>> the
>> > model\n");
>> > +        return AVERROR(EIO);
>> > +    }
>> > +
>> > +    return 0;
>> > +}
>> > +
>> > +static int config_output(AVFilterLink *outlink)
>> > +{
>> > +    AVFilterContext *context = outlink->src;
>> > +    DnnProcessingContext *ctx = context->priv;
>> > +    DNNReturnType result;
>> > +
>> > +    // have a try run in case that the dnn model resize the frame
>> > +    result = (ctx->dnn_module->execute_model)(ctx->model,
>> > &ctx->output,
>> 1);
>> > +    if (result != DNN_SUCCESS){
>> > +        av_log(ctx, AV_LOG_ERROR, "failed to execute model\n");
>> > +        return AVERROR(EIO);
>> > +    }
>> > +
>> > +    outlink->w = ctx->output.width;
>> > +    outlink->h = ctx->output.height;
>> > +
>> > +    return 0;
>> > +}
>> > +
>> > +static int copy_from_frame_to_dnn(DNNData *dnn_data, const AVFrame
>> *in)
>> > +{
>> > +    // extend this function to support more formats
>> > +    av_assert0(in->format == AV_PIX_FMT_RGB24 || in->format ==
>> > AV_PIX_FMT_RGB24);
>> > +
>> > +    if (dnn_data->dt == DNN_FLOAT) {
>> > +        float *dnn_input = dnn_data->data;
>> > +        for (int i = 0; i < in->height; i++) {
>> > +            for(int j = 0; j < in->width * 3; j++) {
>> > +                int k = i * in->linesize[0] + j;
>> > +                int t = i * in->width * 3 + j;
>> > +                dnn_input[t] = in->data[0][k] / 255.0f;
>> > +            }
>> > +        }
>> > +    } else {
>> > +        uint8_t *dnn_input = dnn_data->data;
>> > +        av_assert0(dnn_data->dt == DNN_UINT8);
>> > +        for (int i = 0; i < in->height; i++) {
>> > +            for(int j = 0; j < in->width * 3; j++) {
>> > +                int k = i * in->linesize[0] + j;
>> > +                int t = i * in->width * 3 + j;
>> > +                dnn_input[t] = in->data[0][k];
>> > +            }
>> > +        }
>> > +    }
>> > +
>> > +    return 0;
>> > +}
>> > +
>> > +static int copy_from_dnn_to_frame(AVFrame *out, const DNNData
>> *dnn_data)
>> > +{
>> > +    // extend this function to support more formats
>> > +    av_assert0(out->format == AV_PIX_FMT_RGB24 || out->format ==
>> > AV_PIX_FMT_RGB24);
>> > +
>> > +    if (dnn_data->dt == DNN_FLOAT) {
>> > +        float *dnn_output = dnn_data->data;
>> > +        for (int i = 0; i < out->height; i++) {
>> > +            for(int j = 0; j < out->width * 3; j++) {
>> > +                int k = i * out->linesize[0] + j;
>> > +                int t = i * out->width * 3 + j;
>> > +                out->data[0][k] = av_clip((int)(dnn_output[t] *
>> > 255.0f), 0,
>> > 255);
>> > +            }
>> > +        }
>> > +    } else {
>> > +        uint8_t *dnn_output = dnn_data->data;
>> > +        av_assert0(dnn_data->dt == DNN_UINT8);
>> > +        for (int i = 0; i < out->height; i++) {
>> > +            for(int j = 0; j < out->width * 3; j++) {
>> > +                int k = i * out->linesize[0] + j;
>> > +                int t = i * out->width * 3 + j;
>> > +                out->data[0][k] = dnn_output[t];
>> > +            }
>> > +        }
>> > +    }
>> > +
>> > +    return 0;
>> > +}
>> > +
>> > +static int filter_frame(AVFilterLink *inlink, AVFrame *in)
>> > +{
>> > +    AVFilterContext *context  = inlink->dst;
>> > +    AVFilterLink *outlink = context->outputs[0];
>> > +    DnnProcessingContext *ctx = context->priv;
>> > +    DNNReturnType dnn_result;
>> > +    AVFrame *out;
>> > +
>> > +    copy_from_frame_to_dnn(&ctx->input, in);
>> > +
>> > +    dnn_result = (ctx->dnn_module->execute_model)(ctx->model,
>> &ctx->output,
>> > 1);
>> > +    if (dnn_result != DNN_SUCCESS){
>> > +        av_log(ctx, AV_LOG_ERROR, "failed to execute model\n");
>> > +        av_frame_free(&in);
>> > +        return AVERROR(EIO);
>> > +    }
>> > +    av_assert0(ctx->output.channels == 3);
>> > +
>> > +    out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
>> > +    if (!out) {
>> > +        av_log(ctx, AV_LOG_ERROR, "could not allocate memory for
>> output
>> > frame\n");
>>
>> This log message should be removed, as it is not useful at all.
>
> thanks, will remove
>
>>
>> > +        av_frame_free(&in);
>> > +        return AVERROR(ENOMEM);
>> > +    }
>> > +
>> > +    av_frame_copy_props(out, in);
>> > +    copy_from_dnn_to_frame(out, &ctx->output);
>> > +    av_frame_free(&in);
>> > +    return ff_filter_frame(outlink, out);
>> > +}
>> > +
>> > +static av_cold void uninit(AVFilterContext *ctx)
>> > +{
>> > +    DnnProcessingContext *context = ctx->priv;
>> > +
>> > +    if (context->dnn_module)
>> > +        (context->dnn_module->free_model)(&context->model);
>> > +
>> > +    av_freep(&context->dnn_module);
>> > +}
>> > +
>> > +static const AVFilterPad dnn_processing_inputs[] = {
>> > +    {
>> > +        .name         = "default",
>> > +        .type         = AVMEDIA_TYPE_VIDEO,
>> > +        .config_props = config_input,
>> > +        .filter_frame = filter_frame,
>> > +    },
>> > +    { NULL }
>> > +};
>> > +
>> > +static const AVFilterPad dnn_processing_outputs[] = {
>> > +    {
>> > +        .name = "default",
>> > +        .type = AVMEDIA_TYPE_VIDEO,
>> > +        .config_props  = config_output,
>> > +    },
>> > +    { NULL }
>> > +};
>> > +
>> > +AVFilter ff_vf_dnn_processing = {
>> > +    .name          = "dnn_processing",
>> > +    .description   = NULL_IF_CONFIG_SMALL("Apply DNN processing
>> filter to
>> > the input."),
>> > +    .priv_size     = sizeof(DnnProcessingContext),
>> > +    .init          = init,
>> > +    .uninit        = uninit,
>> > +    .query_formats = query_formats,
>> > +    .inputs        = dnn_processing_inputs,
>> > +    .outputs       = dnn_processing_outputs,
>> > +    .priv_class    = &dnn_processing_class,
>> > +    .flags         = AVFILTER_FLAG_SUPPORT_TIMELINE_GENERIC,
>>
>> If filter changes w/h, this can not be supported.
>
> got it, will remove this line, thank you.
>
>>
>> > +};
>> > --
>> > 2.7.4
>> >
>> > _______________________________________________
>> > ffmpeg-devel mailing list
>> > ffmpeg-devel at ffmpeg.org
>> > https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>> >
>> > To unsubscribe, visit link above, or email
>> > ffmpeg-devel-request at ffmpeg.org with subject "unsubscribe".
>


More information about the ffmpeg-devel mailing list