Recherche avancée

Sur d’autres sites (5280)

• Python pipe ffmpeg stream to numpy in realtime

  29 mai 2020, par Dadidum

  im trying to get the frames out of a rtp stream via ffmpeg in realtime in python. At my server the frames are piped from numpy arrays into ffmpeg and should be at the client piped back to numpy for further manipulation. The server creates a sdp file which is given as input to the client. 
It looks like this :

  



  v=0
o=- 0 0 IN IP4 127.0.0.1
s=No Name
c=IN IP4 127.0.0.1
t=0 0
a=tool:libavformat 58.42.102
m=video 5004 RTP/AVP 96
b=AS:19000
a=rtpmap:96 H265/90000


  



  With ffplay or VLC the stream can be received but with my python code not, at first needs ca. 5sec for each frame and the image is just some random coloured pixels.
The client :

  



  dimension = '{}x{}'.format(480, 240)
sdf_file = "video.sdp"

command = ['FFMPEG',
           '-protocol_whitelist', 'udp,rtp,file,pipe,crypto,data',
           '-i', sdf_file,
           '-pix_fmt', 'bgr24',
           '-r', '24',
           '-video_size', dimension,
           '-f', 'image2pipe', '-']

size = 480 * 240 * 3
proc = sp.Popen(command, stdout=sp.PIPE)

while True:
    try:
        tic = time.perf_counter()
        frame = proc.stdout.read(size)
        print(frame)
        if frame is None:
            print('no input')
            break
        image = np.frombuffer(frame, dtype='uint8').reshape(240, 480, 3)
        toc = time.perf_counter()
        print(f"performed calc in {(toc - tic) * 1000:0.4f} miliseconds")
        cv2.imshow('received', image)
    except Exception as e:
        print(e)

cv2.destroyAllWindows()
proc.stdin.close()
proc.wait()


  



  Any ideas, how i can get the frames in realtime ? Thanky for your help

  


• Passing python's file like object to ffmpeg via subprocess

  3 juillet 2012, par tejinderss

  I have a django FileField, which i use to store wav files on the Amazon s3 server. I have set up the celery task to read that file and convert it to mp3 and store it to another FileField. Problem i am facing is that i am unable to pass the input file to ffmpeg as the file is not the physical file on the hard disk drive. To circumvent that, i used stdin to feed the input stream of the file with the django's filefield. Here is the example :

  output_file = NamedTemporaryFile(suffix='.mp3')
  
  subprocess.call(['ffmpeg', '-y', '-i', '-', output_file.name], stdin=recording_wav)

  where recording_wav file is : , which is actually stored on the amazon s3 server.
  The error for the above subprocess call is :

  AttributeError: 'cStringIO.StringO' object has no attribute 'fileno'

  How can i do this ? Thanks in advance for the help.

  Edit :

  Full traceback :

  [2012-07-03 04:09:50,336: ERROR/MainProcess] Task api.tasks.convert_audio[b7ab4192-2bff-4ea4-9421-b664c8d6ae2e] raised exception: AttributeError("'cStringIO.StringO' object has no attribute 'fileno'",)
  
  Traceback (most recent call last):
  
    File "/home/tejinder/envs/tmai/local/lib/python2.7/site-packages/celery/execute/trace.py", line 181, in trace_task
  
      R = retval = fun(*args, **kwargs)
  
    File "/home/tejinder/projects/tmai/../tmai/apps/api/tasks.py", line 56, in convert_audio
  
      subprocess.Popen(['ffmpeg', '-y', '-i', '-', output_file.name], stdin=recording_wav)
  
    File "/usr/lib/python2.7/subprocess.py", line 672, in __init__
  
      errread, errwrite) = self._get_handles(stdin, stdout, stderr)
  
    File "/usr/lib/python2.7/subprocess.py", line 1043, in _get_handles
  
      p2cread = stdin.fileno()
  
    File "/home/tejinder/envs/tmai/local/lib/python2.7/site-packages/django/core/files/utils.py", line 12, in <lambda>
  
      fileno = property(lambda self: self.file.fileno)
  
    File "/home/tejinder/envs/tmai/local/lib/python2.7/site-packages/django/core/files/utils.py", line 12, in <lambda>
  
      fileno = property(lambda self: self.file.fileno)
  
  AttributeError: &#39;cStringIO.StringO&#39; object has no attribute &#39;fileno&#39;
  
  </lambda></lambda>

• Revision 3606b78108 : Modified test for auto key frame detection. The existing test was triggering a

  16 avril 2015, par paulwilkins

  Changed Paths :
   Modify /vp9/encoder/vp9_firstpass.c

  
  
  Modified test for auto key frame detection.

  The existing test was triggering a lot of false positives on some types
  of animated material with very plain backgrounds. These were triggering
  code designed to catch key frames in letter box format clips.

  This patch tightens up the criteria and imposes a minimum requirement
  on the % blocks coded intra in the first pass and the ratio between the
  % coded intra and the modified inter % after discounting neutral (flat)
  blocks that are coded equally well either way.

  On a particular problem animation clip this change eliminated a large
  number of false positives including some cases where the old code
  selected kf several times in a row. Marginal false negatives are less
  damaging typically to compression and in the problem clip there are now
  a couple of cases where "visual" scene cuts are ignored because of well
  correlated content across the scene cut.

  Replaced some magic numbers related to this with #defines and added
  explanatory comments.

  Change-Id : Ia3d304ac60eb7e4323e3817eaf83b4752cd63ecf