cfd-reader

A python3 module to index, pre-process, and supply facial data from the Chicago Face Database (CFD).

A convenient way to load data from the Chicago Face Database, wrapped in numpy arrays, to feed to your neural network.

installation

From a terminal, run the command:

[sudo -H] pip3 install cfd-reader

usage

In your Python (3.5+) script or interpreter session, run (suggested syntax):

import cfd_reader as cfdr

load face data

cfdr.load_data([options])

where options include:

• grayscale=[True, False]:
  • True: provides 3 channel BGR output
  • False: provides single channel grayscale output
• train_proportion=x:
  • x: real-valued variable between 0 and 1. x proportion of all data will be supplied as training data, and 1-x data will be supplied as test data
• resize=[False, (x,y)]:
  • False: keeps images in their original resolution
  • (x,y): tuple of integers to resize the image to

returns:

numpy array of shape (n, x, y, c) where n are the total number of images in that particular set (train/test), (x,y) is the 2D image shape, and c is the number of channels. In other words, (x,y,c) is the image shape.