image-preprocessor 0.0.1

Image preprocessor for convolutional neural networks

Image Preprocessor

Simple image preprocessor I made while working on my first Keras binary classification convolutional neural network. Converts images from specified directories into NumPy arrays. Its functionality will be expanded out of necessity.

Installation

Install through pip as shown:

pip install image-preprocessor

Example usage

from image_preprocessor import ImagePreprocessor

# Vehicle-label correlations
vehicle_types = {
    0 : 'boat',
    1 : 'car',
    2 : 'motorcycle',
    3 : 'plane'
}

# Create the ImagePreprocessor object
ip = ImagePreprocessor(
    pixels=64,
    normalization=255,
    training_threshold=0.7,
    resize_method='square resize',
    color_mode='L'
)

# Prepare the images in select directories
package = ip.preprocess_dirs(
    paths=['images/boat', 'images/car', 'images/motorcycle', 'images/plane'],
    labels=[0, 1, 2, 3],
    partition=True
)

# Go on and do your neural network stuff 
train_features = package['TRAIN_IMAGES']
train_labels = package['TRAIN_LABELS']
test_features = package['TEST_IMAGES']
test_labels = package['TEST_LABELS']