First Convolutional Neural Network Project – Fashion MNIST Classification

First Convolutional Neural Network Project – Fashion MNIST Classification

by Indian AI Production / On / In Deep Learning Projects

Course name: “Machine Learning & Data Science – Beginner to Professional Hands-on Python Course in Hindi”

In the Machine Learning/Data Science/Deep Learning End to End Project in Python Tutorial in Hindi, we explained each and every step of Machine Learning Project / Data Science Project / Deep Learning Project in detail.

Project name: Fashion MNIST Classification using Convolutional Neural Network

What we cover in this Project:

  1. What is Convolutional Neural Network?
  2. What is Google Colab?
  3. Import Libraries
  4. Load Data
  5. Show Image from Numbers
  6. Change Dimension / Feature Scaling
  7. Build First Convolutional Neural Network
  8. Train Model
  9. Test & Evaluate Model
  10. Confusion Matrix
  11. Classification Report
  12. Save Mode
  13. Build 2 Complex CNN

1. What is Convolutional Neural Network?

Image Kernels >>> https://setosa.io/ev/image-kernels/

Hand written App >>> https://www.cs.ryerson.ca/~aharley/vis/conv/

2. What is Google Colab?

Google Colab >>> Click Here

3. Data Preprocessing

  1. Import Libraries
  2. Load Data
  3. Show Image from Numbers
  4. Change Dimension / Feature Scaling
# -*- coding: utf-8 -*-
"""CNN Project - Fashion-MNIST .ipynb

Automatically generated by Colaboratory.

Original file is located at
    https://colab.research.google.com/drive/1KrIrPLKKRwYobmqpacbf9eiQp5UC9Hhu

# Building CNN - Convolutional Neural Network

###Project Fashion-Classification - End to End 

Train Convolutional Neural Network on 60,000 Fashion-MNIST Images (data in NP array)

Test Convolutional Neural Network on 10,000 Fashion-MNIST Images (data in NP array)

## Import Libraries
"""

import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns

import keras # to build Neural Network

"""## Load Data"""

(X_train, y_train), (X_test, y_test) = keras.datasets.fashion_mnist.load_data() # load dataset from  keras

# Print shape of Data

X_train.shape, y_train.shape, "******", X_test.shape, y_test.shape

X_train[0] # image data in 2d numpy array shape 28x28 pixel

y_train[0] #9 => Ankle boot

class_labels = ["T-shirt/top","Trouser","Pullover","Dress","Coat","Sandal","Shirt","Sneaker","Bag","Ankle boot"]
'''
0 => T-shirt/top 
1 => Trouser 
2 => Pullover 
3 => Dress 
4 => Coat 
5 => Sandal 
6 => Shirt 
7 => Sneaker 
8 => Bag 
9 => Ankle boot '''

"""## Show image"""

plt.imshow(X_train[0], cmap='Greys')

plt.figure(figsize=(16,16))

j=1
for i in np.random.randint(0, 1000, 25):
  plt.subplot(5,5,j); j+=1
  plt.imshow(X_train[i], cmap="Greys")
  plt.axis('off') # off the axis
  plt.title('{} / {}'.format(class_labels[y_train[i]], y_train[i]))

"""## Change Dimention"""

X_train.shape

X_train.ndim

# expected conv2d_input to have 4 dimensions, but got array with shape (28, 28, 1)
# so we have increase the dimention 3 to 4
X_train = np.expand_dims(X_train, -1)
X_test = np.expand_dims(X_test, -1)

# ref: https://numpy.org/doc/stable/reference/generated/numpy.expand_dims.html

X_train.ndim

"""## Feature Scaling"""

X_train = X_train/255
X_test = X_test/255

"""## Split Dataset"""

from sklearn.model_selection import train_test_split 
X_train, X_validation, y_train, y_validation = train_test_split(X_train, y_train, test_size= 0.2, random_state=2020)

X_train.shape,  y_train.shape, X_validation.shape, y_validation.shape

4. Build First Convolutional Neural Network

"""# Convolutional Neural Network - Model Building"""

#Building CNN model
cnn_model = keras.models.Sequential([
                         keras.layers.Conv2D(filters=32, kernel_size=3, strides=(1,1), padding='valid',activation= 'relu', input_shape=[28,28,1]),
                         keras.layers.MaxPooling2D(pool_size=(2,2)),
                         keras.layers.Flatten(),
                         keras.layers.Dense(units=128, activation='relu'),
                         keras.layers.Dense(units=10, activation='softmax')
])

cnn_model.summary() # get the summary of model

# complie the model
cnn_model.compile(optimizer='adam', loss= 'sparse_categorical_crossentropy', metrics=['accuracy'])

# train cnn model
cnn_model.fit(X_train, y_train, epochs=10, batch_size=512, verbose=1, validation_data=(X_validation, y_validation))

5. Test and Evaluate Model

  1. Test & Evaluate Model
  2. Confusion Matrix
  3. Classification Report
"""# Test the Model"""

y_pred = cnn_model.predict(X_test)
y_pred.round(2)

y_test

cnn_model.evaluate(X_test, y_test)

"""# Visualize output

plt.figure(figsize=(16,16))

j=1
for i in np.random.randint(0, 1000,25):
  plt.subplot(5,5, j); j+=1
  plt.imshow(X_test[i].reshape(28,28), cmap = 'Greys')
  plt.title('Actual = {} / {} \nPredicted = {} / {}'.format(class_labels[y_test[i]], y_test[i], class_labels[np.argmax(y_pred[i])],np.argmax(y_pred[i])))
  plt.axis('off')
"""

plt.figure(figsize=(16,30))

j=1
for i in np.random.randint(0, 1000,60):
  plt.subplot(10,6, j); j+=1
  plt.imshow(X_test[i].reshape(28,28), cmap = 'Greys')
  plt.title('Actual = {} / {} \nPredicted = {} / {}'.format(class_labels[y_test[i]], y_test[i], class_labels[np.argmax(y_pred[i])],np.argmax(y_pred[i])))
  plt.axis('off')

"""## Confusion Matrix"""

from sklearn.metrics import confusion_matrix

plt.figure(figsize=(16,9))
y_pred_labels = [ np.argmax(label) for label in y_pred ]
cm = confusion_matrix(y_test, y_pred_labels)

# show cm 
sns.heatmap(cm, annot=True, fmt='d',xticklabels=class_labels, yticklabels=class_labels)

from sklearn.metrics import classification_report
cr= classification_report(y_test, y_pred_labels, target_names=class_labels)
print(cr)

6. Save Model

"""# Save Model"""

cnn_model.save('fashion_mnist_cnn_model.h5') # Save model

# Load model
fashion_mnist_cnn_model = keras.models.load_model('fashion_mnist_cnn_model.h5')

Y_pred_sample = fashion_mnist_cnn_model.predict(np.expand_dims(X_test[0], axis=0)).round(2)
Y_pred_sample

np.argmax(Y_pred_sample[0])

y_test[0]

7. Build 2 Complex CNN

"""# Convolutional Neural Network - Building Complex Model"""

#Building CNN model
cnn_model2 = keras.models.Sequential([
                         keras.layers.Conv2D(filters=32, kernel_size=3, strides=(1,1), padding='valid',activation= 'relu', input_shape=[28,28,1]),
                         keras.layers.MaxPooling2D(pool_size=(2,2)),
                         keras.layers.Conv2D(filters=64, kernel_size=3, strides=(2,2), padding='same', activation='relu'),
                         keras.layers.MaxPooling2D(pool_size=(2,2)),
                         keras.layers.Flatten(),
                         keras.layers.Dense(units=128, activation='relu'),
                         keras.layers.Dropout(0.25),
                         keras.layers.Dense(units=256, activation='relu'),
                         keras.layers.Dropout(0.25),
                         keras.layers.Dense(units=128, activation='relu'),
                         keras.layers.Dense(units=10, activation='softmax')
                         ])

# complie the model
cnn_model2.compile(optimizer='adam', loss= 'sparse_categorical_crossentropy', metrics=['accuracy'])

#Train the Model
cnn_model2.fit(X_train, y_train, epochs=20, batch_size=512, verbose=1, validation_data=(X_validation, y_validation))

cnn_model2.save('fashion_mnist_cnn_model2.h5')

"""######## very complex model"""

#Building CNN model
cnn_model3 = keras.models.Sequential([
                         keras.layers.Conv2D(filters=64, kernel_size=3, strides=(1,1), padding='valid',activation= 'relu', input_shape=[28,28,1]),
                         keras.layers.MaxPooling2D(pool_size=(2,2)),
                         keras.layers.Conv2D(filters=128, kernel_size=3, strides=(2,2), padding='same', activation='relu'),
                         keras.layers.MaxPooling2D(pool_size=(2,2)),
                         keras.layers.Conv2D(filters=64, kernel_size=3, strides=(2,2), padding='same', activation='relu'),
                         keras.layers.MaxPooling2D(pool_size=(2,2)),
                         keras.layers.Flatten(),
                         keras.layers.Dense(units=128, activation='relu'),
                         keras.layers.Dropout(0.25),
                         keras.layers.Dense(units=256, activation='relu'),
                         keras.layers.Dropout(0.5),
                         keras.layers.Dense(units=256, activation='relu'),
                         keras.layers.Dropout(0.25),                        
                         keras.layers.Dense(units=128, activation='relu'),
                         keras.layers.Dropout(0.10),                         
                         keras.layers.Dense(units=10, activation='softmax')
                         ])

# complie the model
cnn_model3.compile(optimizer='adam', loss= 'sparse_categorical_crossentropy', metrics=['accuracy'])

#Train the Model
cnn_model3.fit(X_train, y_train, epochs=50, batch_size=512, verbose=1, validation_data=(X_validation, y_validation))

cnn_model3.save('fashion_mnist_cnn_model3.h5')

cnn_model3.evaluate(X_test, y_test)

#End
Download First CNN Project Source Code

Leave a Reply

Top