Recherche avancée

Sur d’autres sites (11710)

• OpenCV and GoPro - empty frames in VideoCapture stream

  19 mai 2014, par Novatar

  I have a GoPro Hero 3+ (Black) which is connected to a video capture card (AverMedia Game Broadcaster HD). I simply want to get the video stream in OpenCV. With a Logitech Webcam there are no problems. The used code is below.

  VideoCapture cap;
  
  cap.open(0);
  
  
  
  waitKey(300);
  
  
  
  //cap.set(CV_CAP_PROP_FRAME_WIDTH, 1280);
  
  //cap.set(CV_CAP_PROP_FRAME_HEIGHT, 720);
  
  
  
  if (cap.isOpened()){
  
      cout << "Cam identified" << endl;
  
  }
  
  
  
  namedWindow("dst", 1);
  
  
  
  while (1){
  
  
  
  Mat frame;
  
  
  
  if (!cap.read(frame)) {
  
      std::cout << "Unable to read frame from video stream" << std::endl;
  
      continue;
  
  }
  
  
  
  imshow("dst", frame);
  
  
  
  [...]
  
  
  
  }

  With the GoPro the following happens : OpenCV is able to open the VideoCapture ("Cam identified") but can’t read any frames (just a gray screen and the output : "Unable to read frame from video stream"). I also checked this with frame.empty() ;.

  I know that the video capture card works correct because Unity opens a WebCamTexture with the GoPro stream without any issues. I read about codec problems in OpenCv and so I already tried to compile OpenCV with FFMPEG support. Now the recorded MP4-Videos of the GoPro can be displayed but the stream still doesn’t work.

  I use OpenCV 2.48, Windows 7 and Visual Studio 2013.

  ---

  EDIT : Here is the code of libVLC solution :

  struct ctx
  
  {
  
  uint8_t* pixeldata;
  
  std::mutex imagemutex;
  
  };
  
  
  
  static void display(void *data, void *id);
  
  static void unlock(void *data, void *id, void *const *p_pixels);
  
  static void *lock(void *data, void **p_pixels);
  
  
  
  struct ctx ctx;
  
  
  
  libvlc_instance_t *inst;
  
  libvlc_media_player_t *mp;
  
  libvlc_media_t *m;
  
  
  
  int main(int argc, char* argv[])
  
  {
  
      ctx.pixeldata = new uint8_t[1280 * 720 * 3];
  
  
  
      char const *vlc_argv[] =
  
      {
  
          "-vvv",
  
          "--no-audio", /* skip any audio track */
  
          "--no-xlib", /* tell VLC to not use Xlib */
  
      };
  
  
  
      int vlc_argc = sizeof(vlc_argv) / sizeof(*vlc_argv);
  
      inst = libvlc_new(vlc_argc, vlc_argv);
  
  
  
      const char *options[] =
  
      {
  
          ":dshow-vdev=AVerMedia HD Capture",
  
          ":dshow-adev=none",
  
          //":dshow-size=1280x720",
  
          ":dshow-fps=24",
  
          ":dshow-chroma=YUY2",
  
          ":dshow-video-input=1",
  
          ":dshow-video-output=1",
  
          ":dshow-aspect-ratio=16\:9",
  
          ":live-caching=80",
  
          NULL
  
      };
  
  
  
      m = libvlc_media_new_location(inst, "dshow://");
  
      for (const char **opt = options; *opt; opt++)
  
          libvlc_media_add_option(m, *opt);
  
  
  
      mp = libvlc_media_player_new_from_media(m);
  
      libvlc_media_release(m);
  
      libvlc_video_set_callbacks(mp, lock, unlock, display, &ctx);
  
      libvlc_video_set_format(mp, "RV24", 1280, 720, 1280 * 3);
  
      libvlc_media_player_play(mp);
  
  
  
      namedWindow("all", 1);
  
  
  
      Mat frame(720, 1280, CV_8UC3);
  
  
  
      while (1){
  
  
  
          ctx.imagemutex.lock();
  
          memcpy(gesamt.data, ctx.pixeldata, 1280 * 720 * sizeof(uint8_t) * 3);
  
          ctx.imagemutex.unlock();
  
  
  
          imshow("all", gesamt);
  
  
  
          if (waitKey(30) == 27) //wait for 'esc' key press for 30ms. If 'esc' key is pressed, break loop
  
          {
  
              cout << "esc key is pressed by user" << endl;
  
              break;
  
          }
  
  
  
      }
  
  
  
      libvlc_media_player_stop(mp);
  
      libvlc_media_player_release(mp);
  
      libvlc_release(inst);
  
      delete[] ctx.pixeldata;
  
  
  
      return 0;
  
  }
  
  
  
  void display(void *data, void *id){
  
      (void)data;
  
      assert(id == NULL);
  
  }
  
  
  
  void unlock(void *data, void *id, void *const *p_pixels){
  
      struct ctx *ctx = (struct ctx*)data;
  
      ctx->imagemutex.unlock();
  
      assert(id == NULL);
  
  }
  
  
  
  void *lock(void *data, void **p_pixels){
  
      struct ctx *ctx = (struct ctx*)data;
  
      ctx->imagemutex.lock();
  
      *p_pixels = ctx->pixeldata;
  
      return NULL;
  
  }

• frame_thread_encoder : use ref-counting to avoid memcpy of all input frames

  23 décembre 2014, par Hendrik Leppkes

  frame_thread_encoder : use ref-counting to avoid memcpy of all input frames
  Apparently uneeded lock/unlock removed by commiter
  Signed-off-by : Michael Niedermayer <michaelni@gmx.at>
  

  • [DH] libavcodec/frame_thread_encoder.c

• Conversion from Iphone Core Surface RGB Frame into ffmepg AVFarme

  10 mars 2015, par Raghu

  I am trying to convert Core Surface RGB frame buffer(Iphone) to ffmpeg Avfarme to encode into a movie file. But I am not getting the correct video output (video showing colors dazzling not the correct picture)

  I guess there is something wrong with converting from core surface frame buffer into AVFrame.

  Here is my code :

  Surface *surface = [[Surface alloc]initWithCoreSurfaceBuffer:coreSurfaceBuffer];
  
  [surface lock];
  
  unsigned int height = surface.height;
  
  unsigned int width = surface.width;
  
  unsigned int alignmentedBytesPerRow = (width * 4);
  
  if (!readblePixels) {
  
      readblePixels = CGBitmapAllocateData(alignmentedBytesPerRow * height);
  
      NSLog(@"alloced readablepixels");
  
  }
  
  unsigned int bytesPerRow = surface.bytesPerRow;
  
  void *pixels = surface.baseAddress;
  
  for (unsigned int j = 0; j < height; j++) {
  
      memcpy(readblePixels + alignmentedBytesPerRow * j, pixels + bytesPerRow * j, bytesPerRow);
  
  }
  
  
  
  pFrameRGB->data[0] = readblePixels; // I guess here is what I am doing wrong.
  
  pFrameRGB->data[1] = NULL;
  
  pFrameRGB->data[2] = NULL;
  
  pFrameRGB->data[3] = NULL;
  
  
  
  pFrameRGB->linesize[0] = pCodecCtx->width;
  
  pFrameRGB->linesize[1] = 0;
  
  pFrameRGB->linesize[2] = 0;
  
  pFrameRGB->linesize[3] = 0;
  
  
  
  
  
  sws_scale (img_convert_ctx, pFrameRGB->data, pFrameRGB->linesize,
  
         0, pCodecCtx->height,
  
         pFrameYUV->data, pFrameYUV->linesize);   

  Please help me out.

  Thanks,

  Raghu