wyn-keras 0.2.0

Our customized keras package for popular image classifier techniques.

wyn-keras 🎉

A Python package for building and experimenting with Popular Convolutional Neural Networks (CNN) and Vision Transformer (ViT) models using TensorFlow and Keras.

Directory Structure 📁

wyn-keras/
├── pyproject.toml
├── README.md
├── wyn_keras
│   ├── __init__.py
│   ├── vgg16.py
│   ├── resnet.py
│   ├── densenet.py
│   ├── inception.py
│   └── vit.py
├── tests
│   └── __init__.py
└── .gitignore

Installation Instructions (From PIP) 📦

To install the package from PyPI, use the following command:

pip install wyn-keras

For more information, visit the PyPI page.

Installation Instructions (Local) 📦

To install the package and its dependencies, use Poetry:

# Install Poetry if you haven't already
curl -sSL https://install.python-poetry.org | python3 -

# Install the package
poetry install

Usage 🚀

Vision Transformer

The ViT class allows you to create and train Vision Transformer models.

VGG16

The VGG16 class allows you to create and train VGG16 models.

ResNet

The ResNet50 class allows you to create and train ResNet50 models.

DenseNet

The DenseNet121 class allows you to create and train DenseNet121 models.

InceptionV3

The InceptionV3Model class allows you to create and train InceptionV3 models.

Additional Functions (Coming Soon...) 🚧

Stay tuned for more functionalities to be added in the future!

Example Usage 📚

Vision Transformer

import tensorflow as tf
from wyn_keras.vit import ViT

# Load and preprocess the MNIST dataset
(x_train, y_train), (x_test, y_test) = tf.keras.datasets.mnist.load_data()
x_train = x_train[..., tf.newaxis].astype("float32") / 255.0
x_test = x_test[..., tf.newaxis].astype("float32") / 255.0

# Number of classes in MNIST dataset
num_classes = 10

# Create an instance of the ViT class
vit_model = ViT(num_classes=num_classes, input_shape=(28, 28, 1), image_size=28, num_epochs=2)

# Create the ViT model
model = vit_model.create_vit_classifier()

# Train the model
history = vit_model.run_experiment(model, x_train, y_train, x_test, y_test)

# Plot patches
vit_model.plot_patches(x_test)

VGG16

from wyn_keras.vgg16 import VGG16

# Define the input shape and number of classes
input_shape = (32, 32, 3)
num_classes = 10

# Create an instance of the VGG16 class
vgg16_instance = VGG16(input_shape=input_shape, num_classes=num_classes)

# Build the model
model_vgg16 = vgg16_instance.build_model()

# Print the model summary
model_vgg16.summary()

ResNet

from wyn_keras.resnet import ResNet50

# Define the input shape, number of classes, kernel size, filters, and strides
input_shape = (224, 224, 3)
num_classes = 1000
kernel_size = 3
filters = [64, 64, 256]
strides = (2, 2)

# Create an instance of the ResNet50 class
resnet_instance = ResNet50(input_shape=input_shape, num_classes=num_classes, kernel_size=kernel_size, filters=filters, strides=strides)

# Build the model
model_resnet = resnet_instance.build_model()

# Print the model summary
model_resnet.summary()

DenseNet

from wyn_keras.densenet import DenseNet121

# Define the input shape, number of classes, growth rate, and compression factor
input_shape = (224, 224, 3)
num_classes = 1000
growth_rate = 32
compression_factor = 0.5

# Create an instance of the DenseNet121 class
densenet_instance = DenseNet121(input_shape=input_shape, num_classes=num_classes, growth_rate=growth_rate, compression_factor=compression_factor)

# Build the model
model_densenet = densenet_instance.build_model()

# Print the model summary
model_densenet.summary()

InceptionV3

from wyn_keras.inception import InceptionV3Model

# Define the input shape, number of classes, and resize shape
input_shape = (32, 32, 3)
num_classes = 10
resize_shape = (75, 75)

# Create an instance of the InceptionV3Model class
inception_instance = InceptionV3Model(input_shape=input_shape, num_classes=num_classes, resize_shape=resize_shape)

# Build the model
model_inception = inception_instance.build_model()

# Print the model summary
model_inception.summary()

Author ✍️

Yiqiao Yin
Email: [e-mail]
Personal Site: https://www.y-yin.io/