avfilter/vf_dnn_processing: add a generic filter for image proccessing with dnn networks
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.
The following is a python script to halve the value of the first
channel of the pixel. It demos how to setup and execute dnn model
with python+tensorflow. It also generates .pb file which will be
used by ffmpeg.
import tensorflow as tf
import numpy as np
import imageio
in_img = imageio.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)
imageio.imsave("out.bmp", np.squeeze(output))
To do the same thing with ffmpeg:
- 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:fmt=rgb24:dnn_backend=tensorflow -y out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-10-31 08:33:02 +00:00
|
|
|
/*
|
|
|
|
* 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"
|
2020-01-13 11:37:56 +00:00
|
|
|
#include "libavutil/imgutils.h"
|
2020-10-07 02:47:31 +00:00
|
|
|
#include "filters.h"
|
2021-01-26 05:35:30 +00:00
|
|
|
#include "dnn_filter_common.h"
|
avfilter/vf_dnn_processing: add a generic filter for image proccessing with dnn networks
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.
The following is a python script to halve the value of the first
channel of the pixel. It demos how to setup and execute dnn model
with python+tensorflow. It also generates .pb file which will be
used by ffmpeg.
import tensorflow as tf
import numpy as np
import imageio
in_img = imageio.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)
imageio.imsave("out.bmp", np.squeeze(output))
To do the same thing with ffmpeg:
- 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:fmt=rgb24:dnn_backend=tensorflow -y out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-10-31 08:33:02 +00:00
|
|
|
#include "formats.h"
|
|
|
|
#include "internal.h"
|
2020-02-20 07:25:57 +00:00
|
|
|
#include "libswscale/swscale.h"
|
2021-01-07 03:14:10 +00:00
|
|
|
#include "libavutil/time.h"
|
avfilter/vf_dnn_processing: add a generic filter for image proccessing with dnn networks
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.
The following is a python script to halve the value of the first
channel of the pixel. It demos how to setup and execute dnn model
with python+tensorflow. It also generates .pb file which will be
used by ffmpeg.
import tensorflow as tf
import numpy as np
import imageio
in_img = imageio.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)
imageio.imsave("out.bmp", np.squeeze(output))
To do the same thing with ffmpeg:
- 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:fmt=rgb24:dnn_backend=tensorflow -y out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-10-31 08:33:02 +00:00
|
|
|
|
|
|
|
typedef struct DnnProcessingContext {
|
|
|
|
const AVClass *class;
|
2021-01-26 05:35:30 +00:00
|
|
|
DnnContext dnnctx;
|
2020-02-21 12:40:07 +00:00
|
|
|
struct SwsContext *sws_uv_scale;
|
|
|
|
int sws_uv_height;
|
avfilter/vf_dnn_processing: add a generic filter for image proccessing with dnn networks
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.
The following is a python script to halve the value of the first
channel of the pixel. It demos how to setup and execute dnn model
with python+tensorflow. It also generates .pb file which will be
used by ffmpeg.
import tensorflow as tf
import numpy as np
import imageio
in_img = imageio.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)
imageio.imsave("out.bmp", np.squeeze(output))
To do the same thing with ffmpeg:
- 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:fmt=rgb24:dnn_backend=tensorflow -y out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-10-31 08:33:02 +00:00
|
|
|
} DnnProcessingContext;
|
|
|
|
|
2021-01-26 05:35:30 +00:00
|
|
|
#define OFFSET(x) offsetof(DnnProcessingContext, dnnctx.x)
|
avfilter/vf_dnn_processing: add a generic filter for image proccessing with dnn networks
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.
The following is a python script to halve the value of the first
channel of the pixel. It demos how to setup and execute dnn model
with python+tensorflow. It also generates .pb file which will be
used by ffmpeg.
import tensorflow as tf
import numpy as np
import imageio
in_img = imageio.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)
imageio.imsave("out.bmp", np.squeeze(output))
To do the same thing with ffmpeg:
- 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:fmt=rgb24:dnn_backend=tensorflow -y out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-10-31 08:33:02 +00:00
|
|
|
#define FLAGS AV_OPT_FLAG_FILTERING_PARAM | AV_OPT_FLAG_VIDEO_PARAM
|
|
|
|
static const AVOption dnn_processing_options[] = {
|
vf_dnn_processing.c: add dnn backend openvino
We can try with the srcnn model from sr filter.
1) get srcnn.pb model file, see filter sr
2) convert srcnn.pb into openvino model with command:
python mo_tf.py --input_model srcnn.pb --data_type=FP32 --input_shape [1,960,1440,1] --keep_shape_ops
See the script at https://github.com/openvinotoolkit/openvino/tree/master/model-optimizer
We'll see srcnn.xml and srcnn.bin at current path, copy them to the
directory where ffmpeg is.
I have also uploaded the model files at https://github.com/guoyejun/dnn_processing/tree/master/models
3) run with openvino backend:
ffmpeg -i input.jpg -vf format=yuv420p,scale=w=iw*2:h=ih*2,dnn_processing=dnn_backend=openvino:model=srcnn.xml:input=x:output=srcnn/Maximum -y srcnn.ov.jpg
(The input.jpg resolution is 720*480)
Also copy the logs on my skylake machine (4 cpus) locally with openvino backend
and tensorflow backend. just for your information.
$ time ./ffmpeg -i 480p.mp4 -vf format=yuv420p,scale=w=iw*2:h=ih*2,dnn_processing=dnn_backend=tensorflow:model=srcnn.pb:input=x:output=y -y srcnn.tf.mp4
…
frame= 343 fps=2.1 q=31.0 Lsize= 2172kB time=00:00:11.76 bitrate=1511.9kbits/s speed=0.0706x
video:1973kB audio:187kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.517637%
[aac @ 0x2f5db80] Qavg: 454.353
real 2m46.781s
user 9m48.590s
sys 0m55.290s
$ time ./ffmpeg -i 480p.mp4 -vf format=yuv420p,scale=w=iw*2:h=ih*2,dnn_processing=dnn_backend=openvino:model=srcnn.xml:input=x:output=srcnn/Maximum -y srcnn.ov.mp4
…
frame= 343 fps=4.0 q=31.0 Lsize= 2172kB time=00:00:11.76 bitrate=1511.9kbits/s speed=0.137x
video:1973kB audio:187kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.517640%
[aac @ 0x31a9040] Qavg: 454.353
real 1m25.882s
user 5m27.004s
sys 0m0.640s
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2020-05-25 08:16:22 +00:00
|
|
|
{ "dnn_backend", "DNN backend", OFFSET(backend_type), AV_OPT_TYPE_INT, { .i64 = 0 }, INT_MIN, INT_MAX, FLAGS, "backend" },
|
avfilter/vf_dnn_processing: add a generic filter for image proccessing with dnn networks
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.
The following is a python script to halve the value of the first
channel of the pixel. It demos how to setup and execute dnn model
with python+tensorflow. It also generates .pb file which will be
used by ffmpeg.
import tensorflow as tf
import numpy as np
import imageio
in_img = imageio.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)
imageio.imsave("out.bmp", np.squeeze(output))
To do the same thing with ffmpeg:
- 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:fmt=rgb24:dnn_backend=tensorflow -y out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-10-31 08:33:02 +00:00
|
|
|
{ "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" },
|
vf_dnn_processing.c: add dnn backend openvino
We can try with the srcnn model from sr filter.
1) get srcnn.pb model file, see filter sr
2) convert srcnn.pb into openvino model with command:
python mo_tf.py --input_model srcnn.pb --data_type=FP32 --input_shape [1,960,1440,1] --keep_shape_ops
See the script at https://github.com/openvinotoolkit/openvino/tree/master/model-optimizer
We'll see srcnn.xml and srcnn.bin at current path, copy them to the
directory where ffmpeg is.
I have also uploaded the model files at https://github.com/guoyejun/dnn_processing/tree/master/models
3) run with openvino backend:
ffmpeg -i input.jpg -vf format=yuv420p,scale=w=iw*2:h=ih*2,dnn_processing=dnn_backend=openvino:model=srcnn.xml:input=x:output=srcnn/Maximum -y srcnn.ov.jpg
(The input.jpg resolution is 720*480)
Also copy the logs on my skylake machine (4 cpus) locally with openvino backend
and tensorflow backend. just for your information.
$ time ./ffmpeg -i 480p.mp4 -vf format=yuv420p,scale=w=iw*2:h=ih*2,dnn_processing=dnn_backend=tensorflow:model=srcnn.pb:input=x:output=y -y srcnn.tf.mp4
…
frame= 343 fps=2.1 q=31.0 Lsize= 2172kB time=00:00:11.76 bitrate=1511.9kbits/s speed=0.0706x
video:1973kB audio:187kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.517637%
[aac @ 0x2f5db80] Qavg: 454.353
real 2m46.781s
user 9m48.590s
sys 0m55.290s
$ time ./ffmpeg -i 480p.mp4 -vf format=yuv420p,scale=w=iw*2:h=ih*2,dnn_processing=dnn_backend=openvino:model=srcnn.xml:input=x:output=srcnn/Maximum -y srcnn.ov.mp4
…
frame= 343 fps=4.0 q=31.0 Lsize= 2172kB time=00:00:11.76 bitrate=1511.9kbits/s speed=0.137x
video:1973kB audio:187kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.517640%
[aac @ 0x31a9040] Qavg: 454.353
real 1m25.882s
user 5m27.004s
sys 0m0.640s
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2020-05-25 08:16:22 +00:00
|
|
|
#endif
|
|
|
|
#if (CONFIG_LIBOPENVINO == 1)
|
|
|
|
{ "openvino", "openvino backend flag", 0, AV_OPT_TYPE_CONST, { .i64 = 2 }, 0, 0, FLAGS, "backend" },
|
avfilter/vf_dnn_processing: add a generic filter for image proccessing with dnn networks
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.
The following is a python script to halve the value of the first
channel of the pixel. It demos how to setup and execute dnn model
with python+tensorflow. It also generates .pb file which will be
used by ffmpeg.
import tensorflow as tf
import numpy as np
import imageio
in_img = imageio.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)
imageio.imsave("out.bmp", np.squeeze(output))
To do the same thing with ffmpeg:
- 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:fmt=rgb24:dnn_backend=tensorflow -y out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-10-31 08:33:02 +00:00
|
|
|
#endif
|
2021-01-26 05:35:30 +00:00
|
|
|
DNN_COMMON_OPTIONS
|
avfilter/vf_dnn_processing: add a generic filter for image proccessing with dnn networks
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.
The following is a python script to halve the value of the first
channel of the pixel. It demos how to setup and execute dnn model
with python+tensorflow. It also generates .pb file which will be
used by ffmpeg.
import tensorflow as tf
import numpy as np
import imageio
in_img = imageio.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)
imageio.imsave("out.bmp", np.squeeze(output))
To do the same thing with ffmpeg:
- 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:fmt=rgb24:dnn_backend=tensorflow -y out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-10-31 08:33:02 +00:00
|
|
|
{ NULL }
|
|
|
|
};
|
|
|
|
|
|
|
|
AVFILTER_DEFINE_CLASS(dnn_processing);
|
|
|
|
|
|
|
|
static av_cold int init(AVFilterContext *context)
|
|
|
|
{
|
|
|
|
DnnProcessingContext *ctx = context->priv;
|
2021-02-07 06:35:22 +00:00
|
|
|
return ff_dnn_init(&ctx->dnnctx, DFT_PROCESS_FRAME, context);
|
avfilter/vf_dnn_processing: add a generic filter for image proccessing with dnn networks
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.
The following is a python script to halve the value of the first
channel of the pixel. It demos how to setup and execute dnn model
with python+tensorflow. It also generates .pb file which will be
used by ffmpeg.
import tensorflow as tf
import numpy as np
import imageio
in_img = imageio.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)
imageio.imsave("out.bmp", np.squeeze(output))
To do the same thing with ffmpeg:
- 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:fmt=rgb24:dnn_backend=tensorflow -y out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-10-31 08:33:02 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static int query_formats(AVFilterContext *context)
|
|
|
|
{
|
2019-12-27 08:34:15 +00:00
|
|
|
static const enum AVPixelFormat pix_fmts[] = {
|
|
|
|
AV_PIX_FMT_RGB24, AV_PIX_FMT_BGR24,
|
vf_dnn_processing: add support for more formats gray8 and grayf32
The following is a python script to halve the value of the gray
image. It demos how to setup and execute dnn model with python+tensorflow.
It also generates .pb file which will be used by ffmpeg.
import tensorflow as tf
import numpy as np
from skimage import color
from skimage import io
in_img = io.imread('input.jpg')
in_img = color.rgb2gray(in_img)
io.imsave('ori_gray.jpg', np.squeeze(in_img))
in_data = np.expand_dims(in_img, axis=0)
in_data = np.expand_dims(in_data, axis=3)
filter_data = np.array([0.5]).reshape(1,1,1,1).astype(np.float32)
filter = tf.Variable(filter_data)
x = tf.placeholder(tf.float32, shape=[1, None, None, 1], 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())
graph_def = tf.graph_util.convert_variables_to_constants(sess, sess.graph_def, ['dnn_out'])
tf.train.write_graph(graph_def, '.', 'halve_gray_float.pb', as_text=False)
print("halve_gray_float.pb generated, please use \
path_to_ffmpeg/tools/python/convert.py to generate halve_gray_float.model\n")
output = sess.run(y, feed_dict={x: in_data})
output = output * 255.0
output = output.astype(np.uint8)
io.imsave("out.jpg", np.squeeze(output))
To do the same thing with ffmpeg:
- generate halve_gray_float.pb with the above script
- generate halve_gray_float.model with tools/python/convert.py
- try with following commands
./ffmpeg -i input.jpg -vf format=grayf32,dnn_processing=model=halve_gray_float.model:input=dnn_in:output=dnn_out:dnn_backend=native out.native.png
./ffmpeg -i input.jpg -vf format=grayf32,dnn_processing=model=halve_gray_float.pb:input=dnn_in:output=dnn_out:dnn_backend=tensorflow out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-12-27 08:34:20 +00:00
|
|
|
AV_PIX_FMT_GRAY8, AV_PIX_FMT_GRAYF32,
|
2020-02-21 06:20:48 +00:00
|
|
|
AV_PIX_FMT_YUV420P, AV_PIX_FMT_YUV422P,
|
|
|
|
AV_PIX_FMT_YUV444P, AV_PIX_FMT_YUV410P, AV_PIX_FMT_YUV411P,
|
2020-12-18 08:08:29 +00:00
|
|
|
AV_PIX_FMT_NV12,
|
2019-12-27 08:34:15 +00:00
|
|
|
AV_PIX_FMT_NONE
|
|
|
|
};
|
2021-08-09 23:25:31 +00:00
|
|
|
return ff_set_common_formats_from_list(context, pix_fmts);
|
2019-12-27 08:34:15 +00:00
|
|
|
}
|
|
|
|
|
vf_dnn_processing: add support for more formats gray8 and grayf32
The following is a python script to halve the value of the gray
image. It demos how to setup and execute dnn model with python+tensorflow.
It also generates .pb file which will be used by ffmpeg.
import tensorflow as tf
import numpy as np
from skimage import color
from skimage import io
in_img = io.imread('input.jpg')
in_img = color.rgb2gray(in_img)
io.imsave('ori_gray.jpg', np.squeeze(in_img))
in_data = np.expand_dims(in_img, axis=0)
in_data = np.expand_dims(in_data, axis=3)
filter_data = np.array([0.5]).reshape(1,1,1,1).astype(np.float32)
filter = tf.Variable(filter_data)
x = tf.placeholder(tf.float32, shape=[1, None, None, 1], 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())
graph_def = tf.graph_util.convert_variables_to_constants(sess, sess.graph_def, ['dnn_out'])
tf.train.write_graph(graph_def, '.', 'halve_gray_float.pb', as_text=False)
print("halve_gray_float.pb generated, please use \
path_to_ffmpeg/tools/python/convert.py to generate halve_gray_float.model\n")
output = sess.run(y, feed_dict={x: in_data})
output = output * 255.0
output = output.astype(np.uint8)
io.imsave("out.jpg", np.squeeze(output))
To do the same thing with ffmpeg:
- generate halve_gray_float.pb with the above script
- generate halve_gray_float.model with tools/python/convert.py
- try with following commands
./ffmpeg -i input.jpg -vf format=grayf32,dnn_processing=model=halve_gray_float.model:input=dnn_in:output=dnn_out:dnn_backend=native out.native.png
./ffmpeg -i input.jpg -vf format=grayf32,dnn_processing=model=halve_gray_float.pb:input=dnn_in:output=dnn_out:dnn_backend=tensorflow out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-12-27 08:34:20 +00:00
|
|
|
#define LOG_FORMAT_CHANNEL_MISMATCH() \
|
|
|
|
av_log(ctx, AV_LOG_ERROR, \
|
|
|
|
"the frame's format %s does not match " \
|
|
|
|
"the model input channel %d\n", \
|
|
|
|
av_get_pix_fmt_name(fmt), \
|
|
|
|
model_input->channels);
|
|
|
|
|
2019-12-27 08:34:15 +00:00
|
|
|
static int check_modelinput_inlink(const DNNData *model_input, const AVFilterLink *inlink)
|
|
|
|
{
|
|
|
|
AVFilterContext *ctx = inlink->dst;
|
|
|
|
enum AVPixelFormat fmt = inlink->format;
|
|
|
|
|
|
|
|
// the design is to add explicit scale filter before this filter
|
|
|
|
if (model_input->height != -1 && model_input->height != inlink->h) {
|
|
|
|
av_log(ctx, AV_LOG_ERROR, "the model requires frame height %d but got %d\n",
|
|
|
|
model_input->height, inlink->h);
|
|
|
|
return AVERROR(EIO);
|
|
|
|
}
|
|
|
|
if (model_input->width != -1 && model_input->width != inlink->w) {
|
|
|
|
av_log(ctx, AV_LOG_ERROR, "the model requires frame width %d but got %d\n",
|
|
|
|
model_input->width, inlink->w);
|
|
|
|
return AVERROR(EIO);
|
|
|
|
}
|
2020-08-28 04:51:44 +00:00
|
|
|
if (model_input->dt != DNN_FLOAT) {
|
2021-01-18 07:56:34 +00:00
|
|
|
avpriv_report_missing_feature(ctx, "data type rather than DNN_FLOAT");
|
2020-08-28 04:51:44 +00:00
|
|
|
return AVERROR(EIO);
|
|
|
|
}
|
avfilter/vf_dnn_processing: add a generic filter for image proccessing with dnn networks
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.
The following is a python script to halve the value of the first
channel of the pixel. It demos how to setup and execute dnn model
with python+tensorflow. It also generates .pb file which will be
used by ffmpeg.
import tensorflow as tf
import numpy as np
import imageio
in_img = imageio.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)
imageio.imsave("out.bmp", np.squeeze(output))
To do the same thing with ffmpeg:
- 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:fmt=rgb24:dnn_backend=tensorflow -y out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-10-31 08:33:02 +00:00
|
|
|
|
2019-12-27 08:34:15 +00:00
|
|
|
switch (fmt) {
|
|
|
|
case AV_PIX_FMT_RGB24:
|
|
|
|
case AV_PIX_FMT_BGR24:
|
|
|
|
if (model_input->channels != 3) {
|
vf_dnn_processing: add support for more formats gray8 and grayf32
The following is a python script to halve the value of the gray
image. It demos how to setup and execute dnn model with python+tensorflow.
It also generates .pb file which will be used by ffmpeg.
import tensorflow as tf
import numpy as np
from skimage import color
from skimage import io
in_img = io.imread('input.jpg')
in_img = color.rgb2gray(in_img)
io.imsave('ori_gray.jpg', np.squeeze(in_img))
in_data = np.expand_dims(in_img, axis=0)
in_data = np.expand_dims(in_data, axis=3)
filter_data = np.array([0.5]).reshape(1,1,1,1).astype(np.float32)
filter = tf.Variable(filter_data)
x = tf.placeholder(tf.float32, shape=[1, None, None, 1], 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())
graph_def = tf.graph_util.convert_variables_to_constants(sess, sess.graph_def, ['dnn_out'])
tf.train.write_graph(graph_def, '.', 'halve_gray_float.pb', as_text=False)
print("halve_gray_float.pb generated, please use \
path_to_ffmpeg/tools/python/convert.py to generate halve_gray_float.model\n")
output = sess.run(y, feed_dict={x: in_data})
output = output * 255.0
output = output.astype(np.uint8)
io.imsave("out.jpg", np.squeeze(output))
To do the same thing with ffmpeg:
- generate halve_gray_float.pb with the above script
- generate halve_gray_float.model with tools/python/convert.py
- try with following commands
./ffmpeg -i input.jpg -vf format=grayf32,dnn_processing=model=halve_gray_float.model:input=dnn_in:output=dnn_out:dnn_backend=native out.native.png
./ffmpeg -i input.jpg -vf format=grayf32,dnn_processing=model=halve_gray_float.pb:input=dnn_in:output=dnn_out:dnn_backend=tensorflow out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-12-27 08:34:20 +00:00
|
|
|
LOG_FORMAT_CHANNEL_MISMATCH();
|
2019-12-27 08:34:15 +00:00
|
|
|
return AVERROR(EIO);
|
|
|
|
}
|
vf_dnn_processing: add support for more formats gray8 and grayf32
The following is a python script to halve the value of the gray
image. It demos how to setup and execute dnn model with python+tensorflow.
It also generates .pb file which will be used by ffmpeg.
import tensorflow as tf
import numpy as np
from skimage import color
from skimage import io
in_img = io.imread('input.jpg')
in_img = color.rgb2gray(in_img)
io.imsave('ori_gray.jpg', np.squeeze(in_img))
in_data = np.expand_dims(in_img, axis=0)
in_data = np.expand_dims(in_data, axis=3)
filter_data = np.array([0.5]).reshape(1,1,1,1).astype(np.float32)
filter = tf.Variable(filter_data)
x = tf.placeholder(tf.float32, shape=[1, None, None, 1], 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())
graph_def = tf.graph_util.convert_variables_to_constants(sess, sess.graph_def, ['dnn_out'])
tf.train.write_graph(graph_def, '.', 'halve_gray_float.pb', as_text=False)
print("halve_gray_float.pb generated, please use \
path_to_ffmpeg/tools/python/convert.py to generate halve_gray_float.model\n")
output = sess.run(y, feed_dict={x: in_data})
output = output * 255.0
output = output.astype(np.uint8)
io.imsave("out.jpg", np.squeeze(output))
To do the same thing with ffmpeg:
- generate halve_gray_float.pb with the above script
- generate halve_gray_float.model with tools/python/convert.py
- try with following commands
./ffmpeg -i input.jpg -vf format=grayf32,dnn_processing=model=halve_gray_float.model:input=dnn_in:output=dnn_out:dnn_backend=native out.native.png
./ffmpeg -i input.jpg -vf format=grayf32,dnn_processing=model=halve_gray_float.pb:input=dnn_in:output=dnn_out:dnn_backend=tensorflow out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-12-27 08:34:20 +00:00
|
|
|
return 0;
|
|
|
|
case AV_PIX_FMT_GRAYF32:
|
2020-02-21 06:20:48 +00:00
|
|
|
case AV_PIX_FMT_YUV420P:
|
|
|
|
case AV_PIX_FMT_YUV422P:
|
|
|
|
case AV_PIX_FMT_YUV444P:
|
|
|
|
case AV_PIX_FMT_YUV410P:
|
|
|
|
case AV_PIX_FMT_YUV411P:
|
2020-12-18 08:08:29 +00:00
|
|
|
case AV_PIX_FMT_NV12:
|
vf_dnn_processing: add support for more formats gray8 and grayf32
The following is a python script to halve the value of the gray
image. It demos how to setup and execute dnn model with python+tensorflow.
It also generates .pb file which will be used by ffmpeg.
import tensorflow as tf
import numpy as np
from skimage import color
from skimage import io
in_img = io.imread('input.jpg')
in_img = color.rgb2gray(in_img)
io.imsave('ori_gray.jpg', np.squeeze(in_img))
in_data = np.expand_dims(in_img, axis=0)
in_data = np.expand_dims(in_data, axis=3)
filter_data = np.array([0.5]).reshape(1,1,1,1).astype(np.float32)
filter = tf.Variable(filter_data)
x = tf.placeholder(tf.float32, shape=[1, None, None, 1], 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())
graph_def = tf.graph_util.convert_variables_to_constants(sess, sess.graph_def, ['dnn_out'])
tf.train.write_graph(graph_def, '.', 'halve_gray_float.pb', as_text=False)
print("halve_gray_float.pb generated, please use \
path_to_ffmpeg/tools/python/convert.py to generate halve_gray_float.model\n")
output = sess.run(y, feed_dict={x: in_data})
output = output * 255.0
output = output.astype(np.uint8)
io.imsave("out.jpg", np.squeeze(output))
To do the same thing with ffmpeg:
- generate halve_gray_float.pb with the above script
- generate halve_gray_float.model with tools/python/convert.py
- try with following commands
./ffmpeg -i input.jpg -vf format=grayf32,dnn_processing=model=halve_gray_float.model:input=dnn_in:output=dnn_out:dnn_backend=native out.native.png
./ffmpeg -i input.jpg -vf format=grayf32,dnn_processing=model=halve_gray_float.pb:input=dnn_in:output=dnn_out:dnn_backend=tensorflow out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-12-27 08:34:20 +00:00
|
|
|
if (model_input->channels != 1) {
|
|
|
|
LOG_FORMAT_CHANNEL_MISMATCH();
|
|
|
|
return AVERROR(EIO);
|
|
|
|
}
|
|
|
|
return 0;
|
2019-12-27 08:34:15 +00:00
|
|
|
default:
|
2021-01-18 07:56:34 +00:00
|
|
|
avpriv_report_missing_feature(ctx, "%s", av_get_pix_fmt_name(fmt));
|
2019-12-27 08:34:15 +00:00
|
|
|
return AVERROR(EIO);
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
avfilter/vf_dnn_processing: add a generic filter for image proccessing with dnn networks
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.
The following is a python script to halve the value of the first
channel of the pixel. It demos how to setup and execute dnn model
with python+tensorflow. It also generates .pb file which will be
used by ffmpeg.
import tensorflow as tf
import numpy as np
import imageio
in_img = imageio.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)
imageio.imsave("out.bmp", np.squeeze(output))
To do the same thing with ffmpeg:
- 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:fmt=rgb24:dnn_backend=tensorflow -y out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-10-31 08:33:02 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static int config_input(AVFilterLink *inlink)
|
|
|
|
{
|
|
|
|
AVFilterContext *context = inlink->dst;
|
|
|
|
DnnProcessingContext *ctx = context->priv;
|
|
|
|
DNNReturnType result;
|
2019-11-22 07:49:57 +00:00
|
|
|
DNNData model_input;
|
2019-12-27 08:34:15 +00:00
|
|
|
int check;
|
avfilter/vf_dnn_processing: add a generic filter for image proccessing with dnn networks
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.
The following is a python script to halve the value of the first
channel of the pixel. It demos how to setup and execute dnn model
with python+tensorflow. It also generates .pb file which will be
used by ffmpeg.
import tensorflow as tf
import numpy as np
import imageio
in_img = imageio.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)
imageio.imsave("out.bmp", np.squeeze(output))
To do the same thing with ffmpeg:
- 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:fmt=rgb24:dnn_backend=tensorflow -y out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-10-31 08:33:02 +00:00
|
|
|
|
2021-01-26 05:35:30 +00:00
|
|
|
result = ff_dnn_get_input(&ctx->dnnctx, &model_input);
|
avfilter/vf_dnn_processing: add a generic filter for image proccessing with dnn networks
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.
The following is a python script to halve the value of the first
channel of the pixel. It demos how to setup and execute dnn model
with python+tensorflow. It also generates .pb file which will be
used by ffmpeg.
import tensorflow as tf
import numpy as np
import imageio
in_img = imageio.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)
imageio.imsave("out.bmp", np.squeeze(output))
To do the same thing with ffmpeg:
- 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:fmt=rgb24:dnn_backend=tensorflow -y out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-10-31 08:33:02 +00:00
|
|
|
if (result != DNN_SUCCESS) {
|
|
|
|
av_log(ctx, AV_LOG_ERROR, "could not get input from the model\n");
|
|
|
|
return AVERROR(EIO);
|
|
|
|
}
|
|
|
|
|
2019-12-27 08:34:15 +00:00
|
|
|
check = check_modelinput_inlink(&model_input, inlink);
|
|
|
|
if (check != 0) {
|
|
|
|
return check;
|
avfilter/vf_dnn_processing: add a generic filter for image proccessing with dnn networks
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.
The following is a python script to halve the value of the first
channel of the pixel. It demos how to setup and execute dnn model
with python+tensorflow. It also generates .pb file which will be
used by ffmpeg.
import tensorflow as tf
import numpy as np
import imageio
in_img = imageio.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)
imageio.imsave("out.bmp", np.squeeze(output))
To do the same thing with ffmpeg:
- 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:fmt=rgb24:dnn_backend=tensorflow -y out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-10-31 08:33:02 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2020-08-28 04:51:44 +00:00
|
|
|
static av_always_inline int isPlanarYUV(enum AVPixelFormat pix_fmt)
|
|
|
|
{
|
|
|
|
const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
|
|
|
|
av_assert0(desc);
|
|
|
|
return !(desc->flags & AV_PIX_FMT_FLAG_RGB) && desc->nb_components == 3;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int prepare_uv_scale(AVFilterLink *outlink)
|
2020-02-20 07:25:57 +00:00
|
|
|
{
|
|
|
|
AVFilterContext *context = outlink->src;
|
|
|
|
DnnProcessingContext *ctx = context->priv;
|
|
|
|
AVFilterLink *inlink = context->inputs[0];
|
|
|
|
enum AVPixelFormat fmt = inlink->format;
|
2020-02-21 12:40:07 +00:00
|
|
|
|
2020-08-28 04:51:44 +00:00
|
|
|
if (isPlanarYUV(fmt)) {
|
2020-02-21 12:40:07 +00:00
|
|
|
if (inlink->w != outlink->w || inlink->h != outlink->h) {
|
2020-12-18 08:08:29 +00:00
|
|
|
if (fmt == AV_PIX_FMT_NV12) {
|
|
|
|
ctx->sws_uv_scale = sws_getContext(inlink->w >> 1, inlink->h >> 1, AV_PIX_FMT_YA8,
|
|
|
|
outlink->w >> 1, outlink->h >> 1, AV_PIX_FMT_YA8,
|
|
|
|
SWS_BICUBIC, NULL, NULL, NULL);
|
|
|
|
ctx->sws_uv_height = inlink->h >> 1;
|
|
|
|
} else {
|
|
|
|
const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(fmt);
|
|
|
|
int sws_src_h = AV_CEIL_RSHIFT(inlink->h, desc->log2_chroma_h);
|
|
|
|
int sws_src_w = AV_CEIL_RSHIFT(inlink->w, desc->log2_chroma_w);
|
|
|
|
int sws_dst_h = AV_CEIL_RSHIFT(outlink->h, desc->log2_chroma_h);
|
|
|
|
int sws_dst_w = AV_CEIL_RSHIFT(outlink->w, desc->log2_chroma_w);
|
|
|
|
ctx->sws_uv_scale = sws_getContext(sws_src_w, sws_src_h, AV_PIX_FMT_GRAY8,
|
|
|
|
sws_dst_w, sws_dst_h, AV_PIX_FMT_GRAY8,
|
|
|
|
SWS_BICUBIC, NULL, NULL, NULL);
|
|
|
|
ctx->sws_uv_height = sws_src_h;
|
|
|
|
}
|
2020-02-21 12:40:07 +00:00
|
|
|
}
|
2020-02-20 07:25:57 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
avfilter/vf_dnn_processing: add a generic filter for image proccessing with dnn networks
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.
The following is a python script to halve the value of the first
channel of the pixel. It demos how to setup and execute dnn model
with python+tensorflow. It also generates .pb file which will be
used by ffmpeg.
import tensorflow as tf
import numpy as np
import imageio
in_img = imageio.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)
imageio.imsave("out.bmp", np.squeeze(output))
To do the same thing with ffmpeg:
- 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:fmt=rgb24:dnn_backend=tensorflow -y out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-10-31 08:33:02 +00:00
|
|
|
static int config_output(AVFilterLink *outlink)
|
|
|
|
{
|
|
|
|
AVFilterContext *context = outlink->src;
|
|
|
|
DnnProcessingContext *ctx = context->priv;
|
|
|
|
DNNReturnType result;
|
2020-08-28 04:51:44 +00:00
|
|
|
AVFilterLink *inlink = context->inputs[0];
|
avfilter/vf_dnn_processing: add a generic filter for image proccessing with dnn networks
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.
The following is a python script to halve the value of the first
channel of the pixel. It demos how to setup and execute dnn model
with python+tensorflow. It also generates .pb file which will be
used by ffmpeg.
import tensorflow as tf
import numpy as np
import imageio
in_img = imageio.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)
imageio.imsave("out.bmp", np.squeeze(output))
To do the same thing with ffmpeg:
- 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:fmt=rgb24:dnn_backend=tensorflow -y out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-10-31 08:33:02 +00:00
|
|
|
|
2020-08-28 04:51:44 +00:00
|
|
|
// have a try run in case that the dnn model resize the frame
|
2021-01-26 05:35:30 +00:00
|
|
|
result = ff_dnn_get_output(&ctx->dnnctx, inlink->w, inlink->h, &outlink->w, &outlink->h);
|
2020-09-11 14:15:04 +00:00
|
|
|
if (result != DNN_SUCCESS) {
|
|
|
|
av_log(ctx, AV_LOG_ERROR, "could not get output from the model\n");
|
vf_dnn_processing: add support for more formats gray8 and grayf32
The following is a python script to halve the value of the gray
image. It demos how to setup and execute dnn model with python+tensorflow.
It also generates .pb file which will be used by ffmpeg.
import tensorflow as tf
import numpy as np
from skimage import color
from skimage import io
in_img = io.imread('input.jpg')
in_img = color.rgb2gray(in_img)
io.imsave('ori_gray.jpg', np.squeeze(in_img))
in_data = np.expand_dims(in_img, axis=0)
in_data = np.expand_dims(in_data, axis=3)
filter_data = np.array([0.5]).reshape(1,1,1,1).astype(np.float32)
filter = tf.Variable(filter_data)
x = tf.placeholder(tf.float32, shape=[1, None, None, 1], 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())
graph_def = tf.graph_util.convert_variables_to_constants(sess, sess.graph_def, ['dnn_out'])
tf.train.write_graph(graph_def, '.', 'halve_gray_float.pb', as_text=False)
print("halve_gray_float.pb generated, please use \
path_to_ffmpeg/tools/python/convert.py to generate halve_gray_float.model\n")
output = sess.run(y, feed_dict={x: in_data})
output = output * 255.0
output = output.astype(np.uint8)
io.imsave("out.jpg", np.squeeze(output))
To do the same thing with ffmpeg:
- generate halve_gray_float.pb with the above script
- generate halve_gray_float.model with tools/python/convert.py
- try with following commands
./ffmpeg -i input.jpg -vf format=grayf32,dnn_processing=model=halve_gray_float.model:input=dnn_in:output=dnn_out:dnn_backend=native out.native.png
./ffmpeg -i input.jpg -vf format=grayf32,dnn_processing=model=halve_gray_float.pb:input=dnn_in:output=dnn_out:dnn_backend=tensorflow out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-12-27 08:34:20 +00:00
|
|
|
return AVERROR(EIO);
|
avfilter/vf_dnn_processing: add a generic filter for image proccessing with dnn networks
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.
The following is a python script to halve the value of the first
channel of the pixel. It demos how to setup and execute dnn model
with python+tensorflow. It also generates .pb file which will be
used by ffmpeg.
import tensorflow as tf
import numpy as np
import imageio
in_img = imageio.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)
imageio.imsave("out.bmp", np.squeeze(output))
To do the same thing with ffmpeg:
- 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:fmt=rgb24:dnn_backend=tensorflow -y out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-10-31 08:33:02 +00:00
|
|
|
}
|
|
|
|
|
2020-08-28 04:51:44 +00:00
|
|
|
prepare_uv_scale(outlink);
|
avfilter/vf_dnn_processing: add a generic filter for image proccessing with dnn networks
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.
The following is a python script to halve the value of the first
channel of the pixel. It demos how to setup and execute dnn model
with python+tensorflow. It also generates .pb file which will be
used by ffmpeg.
import tensorflow as tf
import numpy as np
import imageio
in_img = imageio.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)
imageio.imsave("out.bmp", np.squeeze(output))
To do the same thing with ffmpeg:
- 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:fmt=rgb24:dnn_backend=tensorflow -y out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-10-31 08:33:02 +00:00
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2020-02-21 06:20:48 +00:00
|
|
|
static int copy_uv_planes(DnnProcessingContext *ctx, AVFrame *out, const AVFrame *in)
|
|
|
|
{
|
2020-03-18 04:23:10 +00:00
|
|
|
const AVPixFmtDescriptor *desc;
|
|
|
|
int uv_height;
|
|
|
|
|
2020-02-21 12:40:07 +00:00
|
|
|
if (!ctx->sws_uv_scale) {
|
|
|
|
av_assert0(in->height == out->height && in->width == out->width);
|
2020-03-18 04:23:10 +00:00
|
|
|
desc = av_pix_fmt_desc_get(in->format);
|
|
|
|
uv_height = AV_CEIL_RSHIFT(in->height, desc->log2_chroma_h);
|
2020-02-21 12:40:07 +00:00
|
|
|
for (int i = 1; i < 3; ++i) {
|
|
|
|
int bytewidth = av_image_get_linesize(in->format, in->width, i);
|
2021-05-11 06:05:18 +00:00
|
|
|
if (bytewidth < 0) {
|
|
|
|
return AVERROR(EINVAL);
|
|
|
|
}
|
2020-02-21 12:40:07 +00:00
|
|
|
av_image_copy_plane(out->data[i], out->linesize[i],
|
|
|
|
in->data[i], in->linesize[i],
|
|
|
|
bytewidth, uv_height);
|
|
|
|
}
|
2020-12-18 08:08:29 +00:00
|
|
|
} else if (in->format == AV_PIX_FMT_NV12) {
|
|
|
|
sws_scale(ctx->sws_uv_scale, (const uint8_t **)(in->data + 1), in->linesize + 1,
|
|
|
|
0, ctx->sws_uv_height, out->data + 1, out->linesize + 1);
|
2020-02-21 12:40:07 +00:00
|
|
|
} else {
|
|
|
|
sws_scale(ctx->sws_uv_scale, (const uint8_t **)(in->data + 1), in->linesize + 1,
|
|
|
|
0, ctx->sws_uv_height, out->data + 1, out->linesize + 1);
|
|
|
|
sws_scale(ctx->sws_uv_scale, (const uint8_t **)(in->data + 2), in->linesize + 2,
|
|
|
|
0, ctx->sws_uv_height, out->data + 2, out->linesize + 2);
|
2020-02-21 06:20:48 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2021-01-07 03:14:10 +00:00
|
|
|
static int flush_frame(AVFilterLink *outlink, int64_t pts, int64_t *out_pts)
|
|
|
|
{
|
|
|
|
DnnProcessingContext *ctx = outlink->src->priv;
|
|
|
|
int ret;
|
|
|
|
DNNAsyncStatusType async_state;
|
|
|
|
|
2021-01-26 05:35:30 +00:00
|
|
|
ret = ff_dnn_flush(&ctx->dnnctx);
|
2021-01-07 03:14:10 +00:00
|
|
|
if (ret != DNN_SUCCESS) {
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
do {
|
|
|
|
AVFrame *in_frame = NULL;
|
|
|
|
AVFrame *out_frame = NULL;
|
2021-08-25 21:10:45 +00:00
|
|
|
async_state = ff_dnn_get_result(&ctx->dnnctx, &in_frame, &out_frame);
|
2021-01-07 03:14:10 +00:00
|
|
|
if (out_frame) {
|
|
|
|
if (isPlanarYUV(in_frame->format))
|
|
|
|
copy_uv_planes(ctx, out_frame, in_frame);
|
|
|
|
av_frame_free(&in_frame);
|
|
|
|
ret = ff_filter_frame(outlink, out_frame);
|
|
|
|
if (ret < 0)
|
|
|
|
return ret;
|
|
|
|
if (out_pts)
|
|
|
|
*out_pts = out_frame->pts + pts;
|
|
|
|
}
|
|
|
|
av_usleep(5000);
|
|
|
|
} while (async_state >= DAST_NOT_READY);
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2021-08-25 21:10:46 +00:00
|
|
|
static int activate(AVFilterContext *filter_ctx)
|
2020-12-16 06:07:00 +00:00
|
|
|
{
|
|
|
|
AVFilterLink *inlink = filter_ctx->inputs[0];
|
|
|
|
AVFilterLink *outlink = filter_ctx->outputs[0];
|
2021-01-26 05:35:30 +00:00
|
|
|
DnnProcessingContext *ctx = filter_ctx->priv;
|
2020-12-16 06:07:00 +00:00
|
|
|
AVFrame *in = NULL, *out = NULL;
|
|
|
|
int64_t pts;
|
|
|
|
int ret, status;
|
|
|
|
int got_frame = 0;
|
|
|
|
int async_state;
|
|
|
|
|
|
|
|
FF_FILTER_FORWARD_STATUS_BACK(outlink, inlink);
|
|
|
|
|
|
|
|
do {
|
|
|
|
// drain all input frames
|
|
|
|
ret = ff_inlink_consume_frame(inlink, &in);
|
|
|
|
if (ret < 0)
|
|
|
|
return ret;
|
|
|
|
if (ret > 0) {
|
|
|
|
out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
|
|
|
|
if (!out) {
|
|
|
|
av_frame_free(&in);
|
|
|
|
return AVERROR(ENOMEM);
|
|
|
|
}
|
|
|
|
av_frame_copy_props(out, in);
|
2021-08-25 21:10:45 +00:00
|
|
|
if (ff_dnn_execute_model(&ctx->dnnctx, in, out) != DNN_SUCCESS) {
|
2021-01-07 03:14:10 +00:00
|
|
|
return AVERROR(EIO);
|
2020-12-16 06:07:00 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
} while (ret > 0);
|
|
|
|
|
|
|
|
// drain all processed frames
|
|
|
|
do {
|
|
|
|
AVFrame *in_frame = NULL;
|
|
|
|
AVFrame *out_frame = NULL;
|
2021-08-25 21:10:45 +00:00
|
|
|
async_state = ff_dnn_get_result(&ctx->dnnctx, &in_frame, &out_frame);
|
2020-12-16 06:07:00 +00:00
|
|
|
if (out_frame) {
|
|
|
|
if (isPlanarYUV(in_frame->format))
|
|
|
|
copy_uv_planes(ctx, out_frame, in_frame);
|
|
|
|
av_frame_free(&in_frame);
|
|
|
|
ret = ff_filter_frame(outlink, out_frame);
|
|
|
|
if (ret < 0)
|
|
|
|
return ret;
|
|
|
|
got_frame = 1;
|
|
|
|
}
|
|
|
|
} while (async_state == DAST_SUCCESS);
|
|
|
|
|
|
|
|
// if frame got, schedule to next filter
|
|
|
|
if (got_frame)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
if (ff_inlink_acknowledge_status(inlink, &status, &pts)) {
|
|
|
|
if (status == AVERROR_EOF) {
|
2021-01-07 03:14:10 +00:00
|
|
|
int64_t out_pts = pts;
|
|
|
|
ret = flush_frame(outlink, pts, &out_pts);
|
|
|
|
ff_outlink_set_status(outlink, status, out_pts);
|
2020-12-16 06:07:00 +00:00
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
FF_FILTER_FORWARD_WANTED(outlink, inlink);
|
|
|
|
|
2021-01-07 03:14:10 +00:00
|
|
|
return 0;
|
2020-12-16 06:07:00 +00:00
|
|
|
}
|
|
|
|
|
avfilter/vf_dnn_processing: add a generic filter for image proccessing with dnn networks
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.
The following is a python script to halve the value of the first
channel of the pixel. It demos how to setup and execute dnn model
with python+tensorflow. It also generates .pb file which will be
used by ffmpeg.
import tensorflow as tf
import numpy as np
import imageio
in_img = imageio.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)
imageio.imsave("out.bmp", np.squeeze(output))
To do the same thing with ffmpeg:
- 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:fmt=rgb24:dnn_backend=tensorflow -y out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-10-31 08:33:02 +00:00
|
|
|
static av_cold void uninit(AVFilterContext *ctx)
|
|
|
|
{
|
|
|
|
DnnProcessingContext *context = ctx->priv;
|
|
|
|
|
2020-02-21 12:40:07 +00:00
|
|
|
sws_freeContext(context->sws_uv_scale);
|
2021-01-26 05:35:30 +00:00
|
|
|
ff_dnn_uninit(&context->dnnctx);
|
avfilter/vf_dnn_processing: add a generic filter for image proccessing with dnn networks
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.
The following is a python script to halve the value of the first
channel of the pixel. It demos how to setup and execute dnn model
with python+tensorflow. It also generates .pb file which will be
used by ffmpeg.
import tensorflow as tf
import numpy as np
import imageio
in_img = imageio.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)
imageio.imsave("out.bmp", np.squeeze(output))
To do the same thing with ffmpeg:
- 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:fmt=rgb24:dnn_backend=tensorflow -y out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-10-31 08:33:02 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static const AVFilterPad dnn_processing_inputs[] = {
|
|
|
|
{
|
|
|
|
.name = "default",
|
|
|
|
.type = AVMEDIA_TYPE_VIDEO,
|
|
|
|
.config_props = config_input,
|
|
|
|
},
|
|
|
|
};
|
|
|
|
|
|
|
|
static const AVFilterPad dnn_processing_outputs[] = {
|
|
|
|
{
|
|
|
|
.name = "default",
|
|
|
|
.type = AVMEDIA_TYPE_VIDEO,
|
|
|
|
.config_props = config_output,
|
|
|
|
},
|
|
|
|
};
|
|
|
|
|
2021-04-19 16:33:56 +00:00
|
|
|
const AVFilter ff_vf_dnn_processing = {
|
avfilter/vf_dnn_processing: add a generic filter for image proccessing with dnn networks
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.
The following is a python script to halve the value of the first
channel of the pixel. It demos how to setup and execute dnn model
with python+tensorflow. It also generates .pb file which will be
used by ffmpeg.
import tensorflow as tf
import numpy as np
import imageio
in_img = imageio.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)
imageio.imsave("out.bmp", np.squeeze(output))
To do the same thing with ffmpeg:
- 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:fmt=rgb24:dnn_backend=tensorflow -y out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-10-31 08:33:02 +00:00
|
|
|
.name = "dnn_processing",
|
|
|
|
.description = NULL_IF_CONFIG_SMALL("Apply DNN processing filter to the input."),
|
|
|
|
.priv_size = sizeof(DnnProcessingContext),
|
|
|
|
.init = init,
|
|
|
|
.uninit = uninit,
|
2021-08-12 11:05:31 +00:00
|
|
|
FILTER_INPUTS(dnn_processing_inputs),
|
|
|
|
FILTER_OUTPUTS(dnn_processing_outputs),
|
2021-09-27 10:07:35 +00:00
|
|
|
FILTER_QUERY_FUNC(query_formats),
|
avfilter/vf_dnn_processing: add a generic filter for image proccessing with dnn networks
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.
The following is a python script to halve the value of the first
channel of the pixel. It demos how to setup and execute dnn model
with python+tensorflow. It also generates .pb file which will be
used by ffmpeg.
import tensorflow as tf
import numpy as np
import imageio
in_img = imageio.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)
imageio.imsave("out.bmp", np.squeeze(output))
To do the same thing with ffmpeg:
- 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:fmt=rgb24:dnn_backend=tensorflow -y out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-10-31 08:33:02 +00:00
|
|
|
.priv_class = &dnn_processing_class,
|
2021-08-25 21:10:46 +00:00
|
|
|
.activate = activate,
|
avfilter/vf_dnn_processing: add a generic filter for image proccessing with dnn networks
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.
The following is a python script to halve the value of the first
channel of the pixel. It demos how to setup and execute dnn model
with python+tensorflow. It also generates .pb file which will be
used by ffmpeg.
import tensorflow as tf
import numpy as np
import imageio
in_img = imageio.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)
imageio.imsave("out.bmp", np.squeeze(output))
To do the same thing with ffmpeg:
- 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:fmt=rgb24:dnn_backend=tensorflow -y out.tf.png
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
2019-10-31 08:33:02 +00:00
|
|
|
};
|