Recherche avancée

Sur d’autres sites (13094)

• avcodec/v210dec : add support for frame and slice threading

  25 novembre 2019, par Limin Wang

  avcodec/v210dec : add support for frame and slice threading
  1, Test server configure :
  
  [root@localhost ]# cat /proc/cpuinfo  |grep "model name"
  
   model name : Intel(R) Xeon(R) CPU E5-2650 v2 @ 2.60GHz
  
   model name : Intel(R) Xeon(R) CPU E5-2650 v2 @ 2.60GHz
  
   ...
  [root@localhost ]# free -h
  
                total        used        free      shared  buff/cache   available
  
                Mem :           102G        1.1G        100G         16M        657M        100G
  
                Swap :          4.0G          0B        4.0G
  2, Test result :
  
  encode the v210 input data for testing :
  
  ./ffmpeg -y -i 4k_422.ts  -c:v v210 -vframes 10 test.avi
  master :
  
  ./ffmpeg -y -threads 1 -stream_loop 1000 -i ./test.avi -benchmark -f null -
  
  frame=10010 fps= 60 q=-0.0 Lsize=N/A time=00:38:26.30 bitrate=N/A speed=13.7x
  
  video:5240kB audio:432432kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead : unknown
  
  bench : utime=101.869s stime=66.181s rtime=167.996s
  
  bench : maxrss=186552kB
  patch applied :
  
  ./ffmpeg -y -threads 2 -thread_type slice -stream_loop 1000 -i ./test.avi -benchmark -f null -
  
  frame=10010 fps= 72 q=-0.0 Lsize=N/A time=00:38:26.30 bitrate=N/A speed=16.5x
  
  video:5240kB audio:432432kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead : unknown
  
  bench : utime=103.562s stime=74.858s rtime=139.599s
  
  bench : maxrss=188616kB
  ./ffmpeg -y -threads 2 -thread_type frame -stream_loop 1000 -i ./test.avi -benchmark -f null -
  
  frame=10010 fps= 85 q=-0.0 Lsize=N/A time=00:38:26.30 bitrate=N/A speed=19.6x
  
  video:5240kB audio:432432kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead : unknown
  
  bench : utime=114.310s stime=92.685s rtime=117.693s
  
  bench : maxrss=231896kB
  Signed-off-by : Limin Wang <lance.lmwang@gmail.com>
  
  Signed-off-by : Michael Niedermayer <michael@niedermayer.cc>
  

  • [DH] libavcodec/v210dec.c
  • [DH] libavcodec/v210dec.h

• avfilter/vf_dnn_processing : add a generic filter for image proccessing with dnn networks

  31 octobre 2019, par Guo, Yejun

  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>
  

  • [DH] configure
  • [DH] doc/filters.texi
  • [DH] libavfilter/Makefile
  • [DH] libavfilter/allfilters.c
  • [DH] libavfilter/vf_dnn_processing.c

• avcodec/v410dec : add support for frame and slice threading

  25 novembre 2019, par Limin Wang

  avcodec/v410dec : add support for frame and slice threading
  1, Test server configure :
  
  [root@localhost ]# cat /proc/cpuinfo  |grep "model name"
  
  model name	 : Intel(R) Xeon(R) CPU E5-2650 v2 @ 2.60GHz
  
  model name	 : Intel(R) Xeon(R) CPU E5-2650 v2 @ 2.60GHz
  
  ...
  [root@localhost ]# free -h
  
                total        used        free      shared  buff/cache   available
  
                Mem :           102G        1.1G        100G         16M        657M        100G
  
                Swap :          4.0G          0B        4.0G
  2, Test result :
  
  encode the v410 input data for testing :
  
   ./ffmpeg -y -i 4k_422.ts  -c:v v410 -vframes 10 test.avi
  master :
  
  ./ffmpeg -y -stream_loop 1000 -i ./test.avi -benchmark -f null -
  
  frame=10010 fps= 37 q=-0.0 Lsize=N/A time=00:38:26.30 bitrate=N/A speed= 8.6x
  
  video:5240kB audio:432432kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead : unknown
  
  bench : utime=166.016s stime=102.192s rtime=268.120s
  
  bench : maxrss=273400kB
  patch applied :
  
  ./ffmpeg -y -threads 2 -thread_type slice -stream_loop 1000 -i ./test.avi -benchmark -f null -
  
  frame=10010 fps= 53 q=-0.0 Lsize=N/A time=00:38:26.30 bitrate=N/A speed=12.3x
  
  video:5240kB audio:432432kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead : unknown
  
  bench : utime=165.135s stime=100.456s rtime=187.994s
  
  bench : maxrss=275476kB
  ./ffmpeg -y -threads 2 -thread_type frame -stream_loop 1000 -i ./test.avi -benchmark -f null -
  
  frame=10010 fps= 61 q=-0.0 Lsize=N/A time=00:38:26.30 bitrate=N/A speed=14.1x
  
  video:5240kB audio:432432kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead : unknown
  
  bench : utime=171.386s stime=122.102s rtime=163.637s
  
  bench : maxrss=340308kB
  Signed-off-by : Limin Wang <lance.lmwang@gmail.com>
  
  Signed-off-by : Michael Niedermayer <michael@niedermayer.cc>
  

  • [DH] libavcodec/v410dec.c