Recherche avancée

Sur d’autres sites (9810)

• JavaCV Fetch RTSP Meet "avcodec_open2() error -40 : Could not open audio codec" Error

  6 avril 2019, par Jeremy Lu

  I am trying to use javacv to fetch a rtsp monitoring camera and grabber the frames, but meet "avcodec_open2() error -40 : Could not open audio codec" error when the grabber is prepare to start after setup the parameters.

  Javacv version :

  • org.bytedeco.javacv-platform, 1.4.1
  • org.bytedeco.javacpp-presets.opencv-platform, 3.4.1-1.4.1

  Here is the Jave Exception Infomation :

  org.bytedeco.javacv.FrameGrabber$Exception: avcodec_open2() error -40: Could not open audio codec.  
  
    at org.bytedeco.javacv.FFmpegFrameGrabber.startUnsafe(FFmpegFrameGrabber.java:835) ~[javacv-1.4.1.jar:1.4.1]  
  
    at org.bytedeco.javacv.FFmpegFrameGrabber.start(FFmpegFrameGrabber.java:663) ~[javacv-1.4.1.jar:1.4.1]  

  rtsp infomation :
  For the rtsp source, I also tried to download the FFplay and use command

  .\ffplay.exe -rtsp_transport tcp rtsp://admin:DWUUUP@10.193.8.71

  to show the frames from monitoring camera, and it works, and with following information

  • Input #0, rtsp, from ’rtsp ://admin:DWUUUP@10.193.8.71’ :

    • Metadata :
      • title : Media Presentation
    • Duration : N/A, start : 0.000000, bitrate : N/A
      • Stream #0:0 : Video : h264 (Main), yuvj420p(pc, bt709, progressive), 1920x1080 [SAR 1:1 DAR 16:9], 15 fps, 24.17 tbr, 90k tbn, 30 tbc
      • Stream #0:1 : Audio : aac, 16000 Hz, 1 channels, fltp

  • [aac @ 000001fc123be880] Audio object type 0 is not implemented. Update your FFmpeg version to the newest one from Git. If the problem still occurs, it means that your file has a feature which has not been implemented.

  And here is the source code :

  FFmpegFrameGrabber grabber = new FFmpegFrameGrabber ("rtsp://admin:DWUUUP@10.193.8.71");
  
  grabber.setImageHeight(640);
  
  grabber.setImageWidth(360);
  
  grabber.setOption("rtsp_transport", "tcp");
  
  grabber.start(); // Where meet the problem
  
  
  
  Java2DFrameConverter converter = new Java2DFrameConverter();
  
  BufferedImage bufferedImage = converter.convert( grabber.grab() );
  
  ByteArrayOutputStream bateArrayOutputStream = new ByteArrayOutputStream();
  
  ImageIO.write(bufferedImage, "jpeg", bateArrayOutputStream);
  
  byte[] data = bateArrayOutputStream.toByteArray();
  
  bateArrayOutputStream.close();

  I have tested to change the rtsp to a mp4 file, the code can work.

  How can I fix this problem ? Thanks very much !

• Installing "ffmpeg" package from setup.py in Apache Beam pipeline running on Google Cloud Dataflow

  17 avril 2019, par John Allard

  I’m trying to run an Apache Beam pipeline on Google Cloud Dataflow that utilizes FFmpeg to perform transcoding operations. As I understand it, since ffmpeg is not a python package (available through PIP), I need to install it from setup.py using the following lines

  # The output of custom commands (including failures) will be logged in the
  
  # worker-startup log.
  
  CUSTOM_COMMANDS = [
  
      ['apt-get', 'update'],
  
      ['apt-get', 'install', '-y', 'ffmpeg']]

  Unfortunately, this is not working. My pipeline is stalling and when I go to examine the logs I’m seeing this

  

  RuntimeError: Command ['apt-get', 'install', '-y', 'ffmpeg'] failed: exit code: 100

  It appears to be unable to find the package ’ffmpeg’. I’m curious as to why this is - ffmpeg is a standard package that should be available under apt-get.

• Can I use the "stream copy" of ffmpeg in OpenCV with VideoWriter class ?

  14 mai 2019, par DerekG

  Python : 3.6, Ubuntu 18.04, OpenCV 4.1.0

  I have an IP camera that streams video data in H.264 encoding. I would like to take this video stream and save it in a .avi file using Python without any encoding or decoding. Using a command line interface and ffmpeg commands, this is trivial to do.

  mycomputer@home:~$ ffmpeg -i rtsp://username:password@192.168.1.1/?framerate=30.0?streamprofile=defaultcameraprofile -acodec copy -vcodec copy output_file_name.avi 

  I’d like to do a similar thing from within Python using OpenCV. Currently, I am using VideoCapture class to read in each frame, and then using the OpenCV VideoWriter class to write this frame to a file. My issue is that the VideoWriter class requires a four_cc code to specify which codec is to be used during writing, whereas I don’t want to use any codec at all (or more precisely, I want to use the stream copy of FFmpeg, but there is to my knowledge no corresponding four_cc code). OpenCV documentation specifies that if I use the FFmpeg backend API and specify both a codec and framerate of 0, the raw stream should be written to the new file. However, when I do this, no file is created.

  cap = cv2.VideoCapture("rtsp://username:password@192.168.1.1/?framerate=30.0?streamprofile=defaultcameraprofile", cv2.CAP_FFMPEG)
  
  frame_width = int(cap.get(3))
  
  frame_height = int(cap.get(4))
  
  # output file, backend API, four_cc code integer, framerate, frame size tuple
  
  out = cv2.videoWriter("output_file.avi",cv2.CAP_FFMPEG,0,0,(frame_width,frame_height))

  I have tried many combinations of other input parameters to the videoWriter constructor but all of them either re-encode the stream or do not write an output file. For instance, I have tried nearly every combination of framerate 0,30, output file extension .avi, .mp4, .h264, .mkv, None and codec 0, "H.264", "RAW ", "DBI ", " ", "MPEG", "COPY", "NONE".

  out = cv2.videoWriter("output_file.avi",cv2.CAP_FFMPEG,cv2.VideoWriter_fourcc('H','2','6','4'),30,(frame_width,frame_height))

  Is it possible to use the FFmpeg stream copy within OpenCV to save video data without encoding it ? If not, I will change tact to running shell commands from within Python, but I’d prefer the former approach if feasible.