import numpy as np

import pandas as pd

import matplotlib.pyplot as plt

import os

from PIL import Image

import cv2

from time import time

data=[]

labels=[]

height = 30

width = 30

channels = 3

classes = 43

n_inputs = height * width

# following accesses images from the directory Train and puts them into numpy arrays.

# Validation images are taken from within the Train folder itself. The test accuracy and confusion matrix

# Will be done using the images in the Test folder.

start = time() # start timer

for i in range(classes) :

path = "train/{0}/".format(i)

print(path)

Class=os.listdir(path)

for a in Class:

try:

image=cv2.imread(path+a, 0)

size_image = cv2.resize(image, (height, width))

data.append(np.array(size_image))

labels.append(i)

except AttributeError:

print(" ")

Cells=np.array(data)

labels=np.array(labels)

#Randomize the order of the input images

s=np.arange(Cells.shape[0])

np.random.seed(43)

np.random.shuffle(s)

Cells=Cells[s]

labels=labels[s]

X_train=Cells[(int)(0.2*len(labels)):]

X_val=Cells[:(int)(0.2*len(labels))]

X_train=X_train.astype('float32')/255

X_val=X_val.astype('float32')/255 # Normalization

y_train=labels[(int)(0.2*len(labels)):]

y_val=labels[:(int)(0.2*len(labels))]

# # # Up until here it's the same, regardless of the type of network to be used.

#Rec-NN

from keras import Sequential

from keras.layers import Dense, Dropout, LSTM, CuDNNLSTM # long short term memory

model = Sequential()

model.add(LSTM(32, input_shape = X_train.shape[1:], activation = 'relu', return_sequences = True))

model.add(Dropout(0.25))

model.add(LSTM(64, activation = 'relu'))

model.add(Dropout(0.25))

model.add(Dense(64, activation = 'relu'))

model.add(Dropout(0.25))

model.add(Dense(43, activation = 'softmax'))

opt = keras.optimizers.Adam(lr = 0.001, decay = 1e-6)

model.compile(loss='sparse_categorical_crossentropy',

optimizer=opt,

metrics=['accuracy'])

history = model.fit(X_train,

y_train,

epochs=10,

validation_data=(X_val, y_val))

# # # From here on, it's the same regardless of the type of network

plt.figure(0)

plt.plot(history.history['accuracy'],label='training Accuracy')

plt.plot(history.history['val_accuracy'],label='val Accuracy')

plt.title('Accuracy')

plt.xlabel('epochs')

plt.ylabel('accuracy')

plt.legend()

plt.show()

plt.figure(1)

plt.plot(history.history['loss'], label='training loss')

plt.plot(history.history['val_loss'], label='val loss')

plt.title('Loss')

plt.xlabel('epochs')

plt.ylabel('loss')

plt.legend()

plt.show()

# This deals with the images in the Test Folder.

y_test=pd.read_csv('Test.csv')

labels = y_test['Path'].as_matrix()

y_test=y_test['ClassId'].values

data=[]

for f in labels:

image=cv2.imread("{0}".format(f), 0)

size_image = cv2.resize(image, (height, width))

data.append(np.array(size_image))

X_test=np.array(data)

X_test = X_test.astype('float32')/255

pred = model.predict_classes(X_test)

from sklearn.metrics import accuracy_score

from sklearn.metrics import confusion_matrix

from sklearn.metrics import f1_score

print("The accuracy score is: ", accuracy_score(y_test, pred))

print("The confusion matrix is: \n", confusion_matrix(y_test, pred)) # Gives the number of True_positive/False_positives/True_negatives/False_nagatives

print("The F1 Score is: ", f1_score(y_test, pred, average = 'macro'))

end = time()

print("Time taken: ", end - start)