Machine learning app that cuts specific sections from a video once trained

Has a threaded M3U8 playlist streamer which you can pick out which quality stream used. Scripts to generate images from downloaded videos and classify them, and a script which takes a video as a video as input and cuts out defined section.

Setup project structure:

• Run pip install -r requirements.txt in the root folder
• Run python scripts/setup_folders.py to create the project structure

Scripts:

• Classify: used to check whether the model is trained correctly for the images being used
  • arguments:
    • -i (path to image)
    • -m (path to model)
• Cut video: used to cut a video into segments
  • arguments:
    • -s (start time in format minutes:seconds or hours:minutes:seconds)
    • -d (duration of clip in same format as start time)
    • -f (filename path)
    • -g (whether to cut images from section True or False)
    • -c (how many times per second to cut images from section in seconds)
    • -a (whether the images being cut are from ad section or not either True or False)
• Download Video: used to download videos from overwatchleague.com
• Get images: used to cut images from a video
  • arguments:
    • -s (images cut per second in seconds)
    • -a (whether the images are ads or not True or False)
    • -c (whether to choose a video from cuts folder)
• Prediction images: used to cut images used for testing model
  • arguments:
    • -v (path to video file)
    • -s (how many times per second to cut images from section in seconds)
    • -a (whether images are ads or not True or False)
• Video classify: used to cut out ad sections from whole video (main file)
  • arguments:
    • -v (path to video file)
    • -c (how many sections of the video to sample)
    • -m (model to use to classify)
    • -o (offset for section of video sample from beginning)