harmeetrai / image-classification

Image Classification App

In this project, we will build a convolution neural network in Keras with python on a CIFAR-10 dataset. Then, we will train and build our classification model and delopy it using streamlit and heroku. Please check out the full app here: https://image-classify-app.herokuapp.com/.

Load Dataset

We will be using the CIFAR-10 dataset that is already included in the keras datasets library:

from tensorflow.keras.datasets import cifar10
import matplotlib.pyplot as plt

(x_train, y_train), (x_test, y_test) = cifar10.load_data()

Check if images are loaded:

plt.figure(figsize=(20, 10))
for i in range(9):
    plt.subplot(330+1+i)
    plt.imshow(x_train[i])
    plt.title(y_train[i])
    plt.axis('off')

Data Processing

Load Libraries

from keras.models import Sequential
from keras.layers import Dense
from keras.layers import Dropout
from keras.layers import Flatten
from keras.constraints import maxnorm
from keras.optimizers import SGD
from keras.layers.convolutional import Conv2D
from keras.layers.convolutional import MaxPooling2D
from keras.utils import np_utils

Train and Test sets

Convert the pixel values of the dataset to float type and then normalize the dataset

x_train = x_train.astype('float32')
x_test = x_test.astype('float32')

x_train = x_train / 255.0
x_test = x_test / 255.0

One-hot encoding for target classes

y_train = np_utils.to_categorical(y_train)
y_test = np_utils.to_categorical(y_test)

num_classes = y_test.shape[1]

Modeling

Create the sequential model and add the layers

model=Sequential()
model.add(Conv2D(32,(3,3),input_shape=(32,32,3),
    padding='same',activation='relu',
    kernel_constraint=maxnorm(3)))
model.add(Dropout(0.2))
model.add(Conv2D(32,(3,3),activation='relu',padding='same',kernel_constraint=maxnorm(3)))
model.add(MaxPooling2D(pool_size=(2,2)))
model.add(Flatten())
model.add(Dense(512,activation='relu',kernel_constraint=maxnorm(3)))
model.add(Dropout(0.5))
model.add(Dense(num_classes, activation='softmax'))

Optimizer and Compile Model

sgd = SGD(lr=0.01, momentum=0.9, decay=(0.01/25), nesterov=False)

model.compile(loss='categorical_crossentropy', optimizer=sgd, metrics=['accuracy'])

Train Model

model.fit(x_train, y_train, validation_data=(x_test,y_test), epochs=25, batch_size=32)

Check Accuracy

acc = model.evaluate(x_test, y_test, verbose=0)
print('Test accuracy:', acc[1])

model.save('cifar10_model.h5')

Testing Model

Using test.jpg (airplane image), we will test to see if the model works

results = {
    0: 'airplane',
    1: 'automobile',
    2: 'bird',
    3: 'cat',
    4: 'deer',
    5: 'dog',
    6: 'frog',
    7: 'horse',
    8: 'ship',
    9: 'truck'
}

from PIL import Image
import numpy as np
img = Image.open('test.jpg')
img = img.resize((32, 32))
img = np.array(img)
img = img.astype('float32')
img = img / 255.0
img = np.expand_dims(img, axis=0)
pred = model.predict(img)
print(pred, "\n",  results[np.argmax(pred)])

Output:

[[9.9999976e-01 6.2572963e-10 5.3366811e-08 1.9468609e-10 2.0309849e-07
  3.6830453e-10 5.5412207e-12 2.0695565e-12 8.1430986e-09 1.1587767e-12]] 
 airplane

The Model works!

Building Streamlit App

Using the model we saved called cifar10_model.h5, we will build an interactive app using streamlit. Learn more about streamlit here.

app.py

This will be the main file used to run the app. Please take a look in the github repo.

setup.sh

This is the setup file for heroku:

mkdir -p ~/.streamlit/
echo "\
[server]\n\
headless = true\n\
port = $PORT\n\
enableCORS = false\n\
\n\
" > ~/.streamlit/config.toml

Procfile

web: sh setup.sh && streamlit run app.py

Requirements.txt

You will need a separate requirements.txt file that is dedicated to streamlit:

streamlit
Pillow==8.3.2
tensorflow-cpu
numpy
pytest-shutil

Deploy to Heroku

Now deploy the code to Heroku. This can be done by first deploying to github and linking your github repo to Heroku.

Credit

Special thanks to this blog post for the inspiration: