Skip to content

Latest commit

 

History

History
executable file
·
60 lines (46 loc) · 1.46 KB

README.md

File metadata and controls

executable file
·
60 lines (46 loc) · 1.46 KB
Raw

DimensionFour

Installation

Make sure you have have the pipenv package manager installed

pip3 install pipenv

Install dependencies using pipenv

pipenv shell
pipenv install

Running the code

Command line usage

  1. Enter the environment using:
pipenv shell
  1. Preprocess your video files:
  • Individally as:
python3 -m dimensionfour.preprocess --input dataset/1.mp4
python3 -m dimensionfour.preprocess --input dataset/2.mp4
...
  • Or in bulk as:
python3 -m dimensionfour.preprocess --input dataset/*

The first time you run dimensionfour.preprocess, it will download a pretrained yolov3 model from dropbox. Preprocessing will output a file called <filename>.d4artifact.zip into the current directory which contains metadata about your video

  1. Assemble your .d4artifact.zip files into a final video:
python3 -m dimensionfour.assemble --input *.d4artifact.zip --output summary.avi

This will output a summarized video of all d4artifacts in the current directory as summary.avi

Options

You can filter the summarized video to only include certain object classes using the --filter option: Default: No Filter

python3 -m dimensionfour.assemble --input *.d4artifact.zip --output summary.avi --filter person

You can set the fps of the output using the --fps option: Default: 30

python3 -m dimensionfour.assemble --input *.d4artifact.zip --output summary.avi --fps 10