README.md

ColorWeight

Visualize (and eventually index) images by their dominant colors.

Right now, given an image like this:

You can get this:

or the data it took to make that image, for further processing, indexing, etc.

[
  {
    "relative_volume": 0.4103605514316013,
    "rgb": [ 92, 112, 117 ]
  },
  {
    "relative_volume": 0.3718001060445387,
    "rgb": [ 45, 39, 28 ]
  },
  {
    "relative_volume": 0.14360683987274656,
    "rgb": [ 168, 84, 44 ]
  },
  {
    "relative_volume": 0.07423250265111347,
    "rgb": [ 195, 171, 129 ]
  }
]

See util/color_analysis.py for details and options.

Installation

• Install Pipenv: brew install pipenv

• Install dependencies: pipenv install

• Activate the environment: pipenv shell or use pipenv run <command>,

• Check out the command line interface in cw/utils/colorweight.py (i.e. pipenv run python cw/utils/colorweight.py --help):

   usage: cw/utils/colorweight.py [-h] [-f {png,json} | -o PATH] [-g WxH]
                                  [-c NUMBER]
                                  image
  
   A simple command line utility for analyzing images by their color.
  
   positional arguments:
     image                 The path to an image on the file system or an HTTP(S)
                           URI. If the arguement is a URI and does not appear to
                           resolve to an image (by file extension), an IIIF Image
                           API service is assumed.
  
   optional arguments:
     -h, --help            show this help message and exit
     -f {png,json}, --format {png,json}
                           If --output is not specified, the format to dump to
                           stdout. 'json' (default) or 'png' are supported.
     -o PATH, --output PATH
                           The path for the output file. The format will be
                           determined by the file extenstion. '.json' or '.png'
                           are supported.
     -g WxH, --geometry WxH
                           The width and height of the output image. Ignored if
                           --output is json. (default: 400x100)
     -c NUMBER, --colors NUMBER
                           The number of colors to report, e.g. -c 3 will report
                           the top three colors. If this number is not provided,
                           numbers 1-12 will be tried in order to determine an
                           optimal number. This number of colors will be
                           reported. This can take a long time.
  

  ... or import and play with cw/utils/color_analysis.py