Recherche avancée

Sur d’autres sites (6546)

• dnn : add layer pad which is equivalent to tf.pad

  29 juillet 2019, par Guo, Yejun

  dnn : add layer pad which is equivalent to tf.pad
  the reason to add this layer first is that vf_sr uses it in its
  
  tensorflow model, and the next plan is to update the python script
  
  to convert tf.pad into native model.
  Signed-off-by : Guo, Yejun <yejun.guo@intel.com>
  
  Signed-off-by : Pedro Arthur <bygrandao@gmail.com>
  

  • [DH] libavfilter/dnn/Makefile
  • [DH] libavfilter/dnn/dnn_backend_native_layer_pad.c
  • [DH] libavfilter/dnn/dnn_backend_native_layer_pad.h

• Python OpenCV real-time blurring with saving to output

  10 janvier 2024, par Oleg Novosad

  I have a live video stream via RTSP from my IP camera. I want to blur faces on that stream and output for mobile usage (HLS, H.264, etc). All this should ideally happen in real-time — with the minimum of resources consumed. I plan to deploy this later to some cloud, so the less money I spend on resources the better.

  


  Currently I have a working solution like so :

  


  

  • I capture video using OpenCV

  


  • I update every frame with Gaussian Blur and save it to some folder

  


  • After some amount of frames I create MP4 / AVI / whatever video and make it accessible via HTTP URL

  


  • All of it is running on Django for now

  


  


  I know I am doing something wrong, can someone suggest a better solution ?

  


• How to make video loop properly ?

  28 mai 2019, par woopwoop399

  I want to play this video in a loop https://www.nicovideo.jp/watch/sm16617386 . I want to play an mp4 file in such a way, that whenever it gets to some point in the video (let’s say, 30.3 seconds), it will loop back (to for example 5.85 seconds).

  I tried to add this code in ffplay.c , it didn’t work well enough, I can hear the transition. I guess seeking isn’t fast enough, or audio needs to be looped in an independant way somehow.

  static void video_refresh(void *opaque, double *remaining_time)
  
  {
  
     (original code here...)
  
      time = get_master_clock(is);
  
      if (isnan(time))
  
          time = (double)is->seek_pos / AV_TIME_BASE;
  
      if (time > jump_when) {
  
          stream_seek(is, (int64_t)(6.0 * AV_TIME_BASE), (int64_t)(0.0 * AV_TIME_BASE), 0);
  
      }
  
  }

  My current plan is to just dig into ffmpeg, understand how video and audio decoders work, and savestate/loadstate the decoders.