Zodiac-Signs

In this notebook, I trained a CNN to classify images from the Chinese Zodiac Signs Kaggle database.

Import Datasets

Since the data already hosted in Kaggle, directly added the data.

Load in the training and test data, split the training data into a training and validation set, then create Data Loaders for each of these sets of data. I perform some simple data augmentation by randomly flipping and rotating the given image data. Did this by defining a torchvision transform. This type of data augmentation should add some positional variety to these images, so that when training a model on this data, it will be robust in the face of geometric changes (i.e. it will recognize a ship, no matter which direction it is facing).

Network Architecture used is ResNet 152.

The following layers were implemented.

LINEAR -> ReLU -> Dropout -> Linear -> ReLU -> Linear -> LogSoftmax

Train the Network Number of epochs used to train the model is 20. Further load the Model with the Lowest Validation Loss and test the Trained Network.