# make-dataset

Quickly create multi-class image datasets by scraping images from Google
Images. Great for training/testing Deep Learning models on custom image data.

The images are stored in separate, labelled class folders similar to those
found in popular image datasets such as ImageNet and MSCOCO.


usage: make-dataset.py [-h] [-s SEARCH] [-n NUM_IMAGES] [-d DIRECTORY]

optional arguments:
  -h, --help            show this help message and exit
  -s SEARCH, --search SEARCH
                        comma separated list of image classes to download (ie.
                        "apple,banana,orange")
  -n NUM_IMAGES, --num_images NUM_IMAGES
                        number of images to download (per class)
  -d DIRECTORY, --directory DIRECTORY
                        save directory
