In [1]:
import numpy as np                               #basically an array
import pandas as pd                              #reading and analyze csv
import matplotlib.pyplot as plt                  #data visualisation
import cv2                                       #comp. vision, image processing, uses numpy as images are 2D array(matrices)
import tensorflow as tf                          #creating neural network (collect, build, train, evaluate, predict)
from PIL import Image                            #manipulate images in python
import os                                        #directory control
from sklearn.model_selection import train_test_split
from keras.utils import to_categorical           #one-hot encoding
from keras.models import Sequential, load_model
from keras.layers import Conv2D, MaxPool2D, Dense, Flatten, Dropout
import tqdm                                     #progress bar
import warnings

Using TensorFlow backend.
In [2]:
data = []
labels = []
classes = 43

for i in range(classes):
    path = os.path.join(os.getcwd(),'train',str(i))
    images = os.listdir(path)
    
    for j in images:
        try:
            image = Image.open(path + '\\'+ j)
            image = image.resize((30,30))
            image = np.array(image)
            data.append(image)
            labels.append(i)
        except:
            print("Error loading image")
In [3]:
#Converting lists into numpy arrays bcoz its faster and takes lesser memory
data = np.array(data)
labels = np.array(labels)
In [4]:
print(data.shape, labels.shape)
# Total Images:39209 of size (30*30*3(3 Means Color)) 

(39209, 30, 30, 3) (39209,)
In [5]:
data[39000] #peaking
Out[5]:
array([[[17, 17, 19],
        [18, 18, 19],
        [18, 17, 18],
        ...,
        [20, 19, 23],
        [21, 19, 21],
        [21, 20, 21]],

       [[17, 19, 23],
        [18, 19, 20],
        [19, 19, 20],
        ...,
        [22, 21, 22],
        [24, 23, 24],
        [21, 21, 22]],

       [[20, 24, 26],
        [16, 18, 20],
        [17, 18, 20],
        ...,
        [20, 19, 21],
        [21, 20, 21],
        [19, 19, 19]],

       ...,

       [[20, 20, 25],
        [19, 19, 24],
        [19, 18, 23],
        ...,
        [30, 28, 32],
        [19, 22, 27],
        [18, 19, 24]],

       [[19, 19, 24],
        [17, 17, 22],
        [18, 17, 22],
        ...,
        [22, 26, 31],
        [18, 25, 31],
        [18, 24, 31]],

       [[16, 17, 21],
        [16, 16, 20],
        [17, 16, 21],
        ...,
        [46, 56, 74],
        [43, 54, 73],
        [42, 53, 73]]], dtype=uint8)
In [7]:
labels[4900]
Out[7]:
3
In [8]:
X_train, X_test, y_train, y_test = train_test_split(data, labels, test_size=0.2, random_state=68)

print(X_train.shape, X_test.shape, y_train.shape, y_test.shape)

(31367, 30, 30, 3) (7842, 30, 30, 3) (31367,) (7842,)
In [9]:
#Converting the labels into one hot encoding
y_train = to_categorical(y_train, 43)
y_test = to_categorical(y_test, 43)
print(X_train.shape, X_test.shape, y_train.shape, y_test.shape)

(31367, 30, 30, 3) (7842, 30, 30, 3) (31367, 43) (7842, 43)
In [14]:
y_train[20000]
Out[14]:
array([0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 1., 0., 0., 0., 0., 0., 0.,
       0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,
       0., 0., 0., 0., 0., 0., 0., 0., 0.], dtype=float32)
In [16]:
#Building the model
model = Sequential()
model.add(Conv2D(filters=32, kernel_size=(5,5), activation='relu', input_shape=X_train.shape[1:]))
model.add(Conv2D(filters=32, kernel_size=(5,5), activation='relu'))
model.add(MaxPool2D(pool_size=(2, 2)))
model.add(Dropout(rate=0.25))
model.add(Conv2D(filters=64, kernel_size=(3, 3), activation='relu'))
model.add(Conv2D(filters=64, kernel_size=(3, 3), activation='relu'))
model.add(MaxPool2D(pool_size=(2, 2)))
model.add(Dropout(rate=0.25))
model.add(Flatten())
model.add(Dense(256, activation='relu'))
model.add(Dropout(rate=0.5))
model.add(Dense(43, activation='softmax'))

warnings.filterwarnings("ignore", category=DeprecationWarning)
In [17]:
model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])
In [18]:
model.summary()

Model: "sequential_2"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
conv2d_5 (Conv2D)            (None, 26, 26, 32)        2432      
_________________________________________________________________
conv2d_6 (Conv2D)            (None, 22, 22, 32)        25632     
_________________________________________________________________
max_pooling2d_3 (MaxPooling2 (None, 11, 11, 32)        0         
_________________________________________________________________
dropout_4 (Dropout)          (None, 11, 11, 32)        0         
_________________________________________________________________
conv2d_7 (Conv2D)            (None, 9, 9, 64)          18496     
_________________________________________________________________
conv2d_8 (Conv2D)            (None, 7, 7, 64)          36928     
_________________________________________________________________
max_pooling2d_4 (MaxPooling2 (None, 3, 3, 64)          0         
_________________________________________________________________
dropout_5 (Dropout)          (None, 3, 3, 64)          0         
_________________________________________________________________
flatten_2 (Flatten)          (None, 576)               0         
_________________________________________________________________
dense_3 (Dense)              (None, 256)               147712    
_________________________________________________________________
dropout_6 (Dropout)          (None, 256)               0         
_________________________________________________________________
dense_4 (Dense)              (None, 43)                11051     
=================================================================
Total params: 242,251
Trainable params: 242,251
Non-trainable params: 0
_________________________________________________________________
In [24]:
history = model.fit(X_train, y_train, batch_size=32, epochs=2, validation_data=(X_test, y_test))
#Final trainig of model

warnings.filterwarnings("ignore", category=DeprecationWarning)

Train on 31367 samples, validate on 7842 samples
Epoch 1/2
31367/31367 [==============================] - 177s 6ms/step - loss: 0.2397 - accuracy: 0.9313 - val_loss: 0.0867 - val_accuracy: 0.9758
Epoch 2/2
31367/31367 [==============================] - 177s 6ms/step - loss: 0.2283 - accuracy: 0.9359 - val_loss: 0.0816 - val_accuracy: 0.9768
In [25]:
model.save("Trafic_signs_model.h5")
In [26]:
#plotting graphs for accuracy 
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()
In [27]:
#plotting graphs for loss 
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()
In [28]:
#testing accuracy on test dataset
from sklearn.metrics import accuracy_score
In [29]:
y_test = pd.read_csv('Test.csv')

labels = y_test["ClassId"].values
imgs = y_test["Path"].values

data=[]

for img in imgs:
    image = Image.open(img)
    image = image.resize((30,30))
    data.append(np.array(image))

X_test=np.array(data)

pred = model.predict_classes(X_test)
In [30]:
#Accuracy with the test data
print(accuracy_score(labels, pred))

0.9472684085510689
In [ ]: