The keyframer's intro from biscuitsnake

keyframer's Introduction

Keyframer

A CLI tool to extract and present key frames from videos by sequentially comparing histograms.

Dependencies

Python 3
OpenCV
Pillow

Setup

pip install -r "requirements.txt"

Usage

python keyframer.py INPUT [OPTIONS]

usage: keyframer.py [-h] [-o OUTPUT] [-i INTERVAL] [-f {jpg,png,bmp}]
                    [-s {low,normal,high}] [-p] [-g GRIDSIZE GRIDSIZE]
                    INPUT

positional arguments:
  INPUT                 Input file or directory containing videos to be
                        processed.

optional arguments:
  -h, --help            show this help message and exit
  -o OUTPUT, --output OUTPUT
                        Output directory.
  -i INTERVAL, --interval INTERVAL
                        Get key frames every x seconds rather than through
                        comparison.
  -f {jpg,png,bmp}, --format {jpg,png,bmp}
                        Format of image frames.
  -s {low,normal,high}, --sensitivity {low,normal,high}
                        Sensitivity of comparison (higher is stricter).
                        Defaults to normal.
  -p, --purge           Remove frame and keyframe images to save space during
                        large processes.
  -g GRIDSIZE GRIDSIZE, --gridsize GRIDSIZE GRIDSIZE
                        Grid size of pdf image generated (WIDTH, HEIGHT).
                        Defaults to 3 by 3.

TODO

Potentially use threading to speed up frame r/w, or other solutions
Improve key frame detection
Option to get key frames every x seconds rather than by comparison
Option to only get pdf (avoid storage issues when working with large files or batch processing)
End result formats other than pdf
Explore realtime scene change detection (ie directly grab key frames from a video to avoid writing frames)

Recommend Projects