CNN for Deep Learning Homework

simple-cnn.py

import numpy as np
import pandas as pd
import os
import cv2
from tqdm import tqdm
from keras.models import Sequential
from keras.layers import Dense, Dropout, Flatten, Convolution2D, MaxPooling2D
from keras.callbacks import EarlyStopping
from keras.utils import to_categorical


train = pd.read_csv('train.csv')
test = pd.read_csv('test.csv')

TRAIN_PATH = 'train_img/'
TEST_PATH = 'test_img/'

# function to read images as arrays
def read_image(img_path):
    img = cv2.imread(img_path, cv2.IMREAD_COLOR)
    img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
    img = cv2.resize(img, (64,64)) # you can resize to  (128,128) or (256,256)
    return img
    
train_data = []
test_data = []
train_labels = train['label'].values

# Load Images
for img in tqdm(train['image_id'].values):
    train_data.append(read_image(TRAIN_PATH + '{}.png'.format(img)))
    
for img in tqdm(test['image_id'].values):
    test_data.append(read_image(TEST_PATH + '{}.png'.format(img)))
    
# normalize the images
x_train = np.array(train_data, np.float32) / 255.
x_test = np.array(test_data, np.float32) / 255.

# target variable - encoding numeric value
label_list = train['label'].tolist()
Y_train = {k:v+1 for v,k in enumerate(set(label_list))}
y_train = [Y_train[k] for k in label_list]


## keras accepts target variable as a ndarray so that we can set one output neuron per class
y_train = to_categorical(y_train)

## neural net architechture

model = Sequential()
model.add(Convolution2D(32, (3,3), activation='relu', padding='same',input_shape = (64,64,3)))
model.add(Convolution2D(32, (3,3), activation='relu'))
model.add(MaxPooling2D(pool_size=(2,2)))

model.add(Convolution2D(64, (3,3), activation='relu', padding='same'))
model.add(Convolution2D(64, (3,3), activation='relu'))
model.add(MaxPooling2D(pool_size=(2,2)))
model.add(Dropout(0.25))

model.add(Convolution2D(128, (3,3), activation='relu', padding='same'))
model.add(Convolution2D(128, (3,3), activation='relu'))
model.add(MaxPooling2D(pool_size=(2,2)))
model.add(Dropout(0.25))

model.add(Flatten())
model.add(Dense(128, activation='relu'))
model.add(Dense(256, activation='relu'))
model.add(Dropout(0.25))
model.add(Dense(y_train.shape[1], activation='softmax'))
model.compile(loss = 'categorical_crossentropy', optimizer = 'adam', metrics = ['accuracy'])

early_stops = EarlyStopping(patience=2, monitor='val_acc')

model.fit(x_train, y_train, batch_size=32, epochs=25, validation_split=0.3, callbacks=[early_stops])

# make prediction
predictions = model.predict(x_test)
predictions = np.argmax(predictions, axis= 1)

# get predicted labels
y_maps = dict()
y_maps = {v:k for k, v in Y_train.items()}
pred_labels = [y_maps[k] for k in predictions]

# make submission
sub1 = pd.DataFrame({'image_id':test.image_id, 'label':pred_labels})
sub1.to_csv('submission_one.csv', index=False)