Classification-of-Cat-and-Dog using CNN with image augmentation

Classification of images between two class cat and dog using CNN with image augmentation .

We are using image augmentation to increase the amount of training data using augmentation by using from keras.preprocessing.image import ImageDataGenerator.

Accuracy we achived : 77-78 % *We achieved 90-91 % percent accuracy by Transfer Learning using VGG Model.If interested refer :https://github.com/VikasSinghBhadouria/Image-Classification-using-VGG-transfer-learning/tree/master *

Data is very limited and costly in some cases such as medical imagery. To get the best out of it , we are using ImageDataGenerator and creating multiple images for each image we have in our training data set by rotation ,zoon in , shearing , tilting ,rescaling.

fill_mode for method used to decide the color of newly generated pixel in case of shear n zoom in.

datagen = ImageDataGenerator( rotation_range=40, width_shift_range=0.2, height_shift_range=0.2, rescale=1./255, shear_range=0.2, zoom_range=0.2, horizontal_flip=True, fill_mode='nearest')

For details , read Documentation :http://keras.io/preprocessing/image/

It uses data that can be downloaded at: https://www.kaggle.com/c/dogs-vs-cats/data In our setup, we:

• created a data/ folder
• created train/ and validation/ subfolders inside data/
• created cats/ and dogs/ subfolders inside train/ and validation/ We are using mini-data as that will be enough for our purpose . So that we have 1000 training examples for each class, and 450 validation examples for both class. In summary, this is our directory structure:

data/
    train/
        dogs/
            dog001.jpg
            dog002.jpg
            ...
        cats/
            cat001.jpg
            cat002.jpg
            ...
    validation/
        dogs/
            dog001.jpg
            dog002.jpg
            ...
        cats/
            cat001.jpg
            cat002.jpg
            ...

'''

Reference : https://gist.github.com/fchollet/0830affa1f7f19fd47b06d4cf89ed44d