GUI application to help in batch sorting of image data for use with keras classifiers.
Using TensorFlow backend.
Usage: batch_classify.py [OPTIONS]
This script reads a pre-trained keras model and tries to use good guesses
from this model to speed up subsequent manual classification. It parses
all images in a given folder and groups these into similar classes. A
batch of images (default 6x5 grid) which are tagged as the same class is
presented to the user. By pressing any key (except Q) these are accepted
as the given class and moved to the desired location. By selecting one or
more images with the mouse, you can reject these suggestions by the
classifier.
Pressing Q will terminate the program.
Options:
--batchsize INTEGER... Number of simultanuous images to classify (grid
size: 2 numbers)
--modelname TEXT Filename of keras classifier model
--verbose BOOLEAN Verbose output. Print additional debug
information
--unsorted_directory TEXT Files will be bulk classified and MOVED from here
--sorted_directory TEXT Desitation folder for sorted dataset
--help Show this message and exit.
The script relies on the following list of python modules, which are all available through pip
- numpy
- keras
- tensorflow
- opencv
- click