About

A library that helps to convert volumetric video data into image frames with adjustable frame capturing rate functionality.

The main reason why I created this library is to ease the video to image frames conversion process. During video annotation we need frames of an video or a subpart of a video and that too custom frame capture rate, so this problems led to this library development.

How to install?

pip install video2images

Requirements

It will be automatically installed

• tqdm, imageio, imageio-ffmpeg, moviepy

Class Object Argument Description

Note:- The capture_rate if not specified will consider the original video frames per second rate

Frame Capture Rate Explaination

Valid Extensions

For Video

• .mov
• .avi
• .mpg
• .mpeg
• .mp4
• .mkv
• .wmv

For Image

• .jpg, .jpeg
• .png
• .bmp
• .tiff, .tif
• .dicom, .dcm

How to use?

Minimal setting (Basic)

from video2images import Video2Images


Video2Images(video_filepath="--path-to-video-file--",
             out_dir="--path-to-output-directory--")

Want to capture frames in between some interval

from video2images import Video2Images


# Lets take start_time to be 2mins i.e 120s
# & end_time to be 10 mins i,e 600s out of 20mins videos

Video2Images(video_filepath="--path-to-video-file--",
             start_time=120,
             end_time=600,
             out_dir="--path-to-output-directory--")

Want only first 'f' frames out of 30 frames per second

from video2images import Video2Images


# Let say f = 20 i.e capturing only first 20 frames
Video2Images(video_filepath="--path-to-video-file--",
             capture_rate=20,
             out_dir="--path-to-output-directory--")

Want output image to be saved in png or other format

from video2images import Video2Images


Video2Images(video_filepath="--path-to-video-file--",
             save_format=".png",
             out_dir="--path-to-output-directory--")

From command line

We can convert video frames to images by using simple command line as follows

$video2images --video_filepath <input video filepath> \
--start_time <some integer number in seconds> \ 
--end_time <some integer number in seconds> \
--capture_rate <some integer capture rate> \
--save_format <any one image extension> \
--out_dir <output folder name where all images will be pushed>

Note:- The output will be saved in a folder named frames_folder_{timestamp}

Output Process Screenshot

Author

I will be happy to connect with you guys!!

Linkedin

Twitter

Citation

@software{abhishek_c_salian_2020_4384148,
  author       = {Abhishek C. Salian},
  title        = {Video2Images: A python library for video frames extraction for video annotation based task},
  month        = dec,
  year         = 2020,
  publisher    = {Zenodo},
  version      = {v1.1},
  doi          = {10.5281/zenodo.4384147},
  url          = {http://doi.org/10.5281/zenodo.4384147}
}

Any suggestions are most welcome.