cwjshen / videoslick

Image extraction and comparison from videos

VideoSlick

Image extraction and comparison from videos

Getting Started

Prerequisites / Libraries Used

This project assumes installation of Python and uses the following third party libraries

• OpenCV - https://opencv.org/
• scikit-image - http://scikit-image.org/docs/dev/api/skimage.html

The OpenCV .pyd file is included in the repository and scikit-image comes included in most Python distributions like Anaconda.

Clone Repository

git clone https://github.com/cwjshen/videoslick.git

Overview of Directories and Files

• extract.py - Script that extracts individual image frames from a given video
• compare.py - Script that compares images using skimage's implementation of Structural Similarity Index
• videos - Directory containing videos to be processed
• golden images - Directory containing golden images by video
• test images - Directory containing grayscaled images by video to compare against golden images.
• output - After running extract.py, an output folder will be generated that will store the extracted images and be marked with a timestamp

The image files are extracted and named as "frame{#}" where # corresponds to the frame position in the video from which it was extracted from.

Running the scripts

extract.py

• extract.py takes one required argument that is the relative path to the target video for image extraction

python extract.py 'videos/testvideo.mp4'

• extract.py also takes some optional arguments:
  • [-h, --help] - Displays help menu for running the script

    python extract.py -h
    

  • [-er EXTRACT_RATE, --extract_rate EXTRACT_RATE] - Time between image extractions in milliseconds. Defaults to 500ms if not provided.

    python extract.py 'videos/testvideo.mp4' -er 2000 
    

  • [-gray, --gray] - Extracts the images in grayscale, used to generate test images

    python extract.py 'videos/testvideo.mp4' -gray
    

    python extract.py 'videos/testvideo.mp4' -er 2000 -gray
    

compare.py

• compare.py takes two required arguments that are:
  • [gold_img_dir] - Relative path to directory containing golden images
  • [test_img_dir] - Relative path to directory containing images to be validated

  python compare.py 'golden images/testvideo' 'test images/testvideo'
  

extractcompare.py

Single script that extracts images from a given video file then compares them with given golden images and outputs a summary of test results

• compare.py takes two required arguments that are:

  • [video path] - Relative path to directory containing golden images
  • [test_img_dir] - Relative path to directory containing images to be validated

• extract.py also takes some optional arguments:

  • [-h, --help] - Displays help menu for running the script

  • [-er EXTRACT_RATE, --extract_rate EXTRACT_RATE] - Time between image extractions in milliseconds. Defaults to 500ms if not provided.

  • [-gray, --gray] - Extracts the images in grayscale, used to generate test images

  python compare.py 'videos/testvideo.mp4' 'golden images/testvideo' -er 2000