Skip to content

Latest commit

 

History

History
33 lines (21 loc) · 1.12 KB

File metadata and controls

33 lines (21 loc) · 1.12 KB

Video to Image at required FPS

This python script converts video to images at a required framerate which can be useful for Extracting Deep Learning training images. There are two files single_convert uses a single video and multi_convert requires the video files specified as an input.

How to use?

Just type in:

python single_convert.py --input videofile.mp4 --output output_dir --frate 1

This will convert the video file into images at 1s per framerate meaning it will extract images every one second.

python multi_convert.py --input videofile.mp4 --output output_dir --frate 1

This will convert the video files into images at 1s per framerate meaning it will extract images every one second.

You can always type in:

python single_convert.py -h or python single_convert.py -h for any help with arguments.

-i --input (input argument)
-o --output (output path)
-fps --framesps (Requried frame rate)
-frate --frate (Required frames per second) default set to 0.1

Requirements

Before running the script, just install open-cv for python using this command:

pip install opencv-python

Happy Extracting images.