Recherche avancée

Sur d’autres sites (5956)

• Fastest way to extract moving dynamic crop from video using ffmpeg

  5 avril 2024, par NateW

  I'm working on an AI project that involves object detection and action recognition on roughly 30 minute videos.

  


  My pipeline is the following :

  


  

  • determine crops using object detection model

  


  • extract crops using Python and write them to disk as individual images for each frame.

  


  • use action recognition model by inputting a series of the crops.

  


  


  The models are fast but actual writing of the crops to disk is slow. Sure, using an SSD would speed it up but I'm sure ffmpeg would greatly speed it up.

  


  Some of the challenges with the crops :

  


  

  • the output size is always 128x128

  


  • the input size is variable

  


  • the crop moves on every frame

  


  


  My process for extracting crops is simple using cv2.imwrite(output_crop_path, crop) in a for loop.

  


  I've done experiments trying to use sndcommand and filter_complex. I tried this https://stackoverflow.com/a/67508233/4447761 but it outputs an image with black below the crop and the image gets wrapped around on the x axis.

  


  

  


• FFmpeg : Concatenated clips extracted using -ss and -t, hang or go out of sync while playing

  20 juillet 2013, par Yogi

  I have a set of movies in different formats, and I am trying to extract small clips from these source movies, and concatenate them into one movie.

  My workflow has been the following :

  1. Convert all the source movies to the same format (width, height, fps, codec)T the scale and pad options are so that all movies oare of the same size, even if their aspect ratio is different.

     fmpeg  -i $infile -vcodec libx264 -strict -2  -vf scale=iw*sar*min(${MAX_WIDTH}/(iw*sar)\,${MAX_HEIGHT}/ih):ih*min(${MAX_WIDTH}/(iw*sar)\,${MAX_HEIGHT}/ih),pad=${MAX_WIDTH}:${MAX_HEIGHT}:(ow-iw)/2:(oh-ih)/2 -b:v 500k -b:a 64k -movflags +faststart -g 10  -r 25 ${outbasename}.mp4

  2. extract clips :

     ffmpeg -ss $starttime -t $duration -i $in_file -vcodec copy -acodec copy $out_file

  3. Then finally combine clips by first making a concat_list.txt file which contains the list of clips to be concatenated, and their duration, and then using ffmpegs's concat demux :

     ffmpeg -f concat -i concat_list.txt -c copy -movflags +faststart $oname

  The problem I am facing is that many of the final videos hang or go out of sync, somewhere in the middle of playing. I have tried using mjpeg as the codec, but I still get the same behavior. I can play the individual extracted clips, and then all seem to play fine in most players. Does anybody know what I am doing wrong ? I am using ffmpeg version 1.2.1.

• Is there a faster way to generate video from pixel arrays using python and ffmpeg ?

  8 mai 2019, par devneal17

  I’ve found a few sources which use python and ffmpeg to generate video from pixel arrays by passing the -f rawvideo flag 1 2. However, this is very slow for high-definition video since each individual pixel must be piped into ffmpeg.

  In fact this is provably wasteful, as I’ve found that 2.5Gb of pixel arrays generates about 80Kb of video. I’ve also chanced upon some examples where javascript can render high quality animations in near-real time 1, which makes me even more suspicious that I’m doing something wrong.

  Is there a way to do this more efficiently, perhaps by piping the differences between pixel arrays into ffmpeg rather than the pixels themselves ?

  (edit) This is the line I’m using. Most executions take the else path that follows.