In [18]:

    

import numpy as np
import pandas as pd
from skimage import io, color, exposure, transform
from sklearn.model_selection import train_test_split
import os
import glob
import h5py

from keras.preprocessing.image import ImageDataGenerator
from keras.models import Sequential, model_from_json
from keras.layers.core import Dense, Dropout, Activation, Flatten, Lambda
from keras.layers.convolutional import Conv2D
from keras.layers.pooling import MaxPooling2D

from keras.optimizers import SGD
from keras.utils import np_utils
from keras.callbacks import LearningRateScheduler, ModelCheckpoint
from keras import backend as K
K.set_image_data_format('channels_first')

from matplotlib import pyplot as plt

NUM_CLASSES = 44
IMG_SIZE = 48


    

In [19]:

    

def preprocess_img(img):
    # Histogram normalization in y
    hsv = color.rgb2hsv(img)
    hsv[:,:,2] = exposure.equalize_hist(hsv[:,:,2])
    img = color.hsv2rgb(hsv)

    # central scrop
    min_side = min(img.shape[:-1])
    centre = img.shape[0]//2, img.shape[1]//2
    img = img[centre[0]-min_side//2:centre[0]+min_side//2,
              centre[1]-min_side//2:centre[1]+min_side//2,
              :]

    # rescale to standard size
    img = transform.resize(img, (IMG_SIZE, IMG_SIZE))

    # roll color axis to axis 0
    img = np.rollaxis(img,-1)

    return img


def get_class(img_path):
    return int(img_path.split('/')[-2])

def cnn_model():
    model = Sequential()

    model.add(Conv2D(32, (3, 3), padding='same',
                     input_shape=(3, IMG_SIZE, IMG_SIZE),
                     activation='relu'))
    model.add(Conv2D(32, (3, 3), activation='relu'))
    model.add(MaxPooling2D(pool_size=(2, 2)))
    model.add(Dropout(0.2))

    model.add(Conv2D(64, (3, 3), padding='same',
                     activation='relu'))
    model.add(Conv2D(64, (3, 3), activation='relu'))
    model.add(MaxPooling2D(pool_size=(2, 2)))
    model.add(Dropout(0.2))

    model.add(Conv2D(128, (3, 3), padding='same',
                     activation='relu'))
    model.add(Conv2D(128, (3, 3), activation='relu'))
    model.add(MaxPooling2D(pool_size=(2, 2)))
    model.add(Dropout(0.2))

    model.add(Flatten())
    model.add(Dense(512, activation='relu'))
    model.add(Dropout(0.5))
    model.add(Dense(NUM_CLASSES, activation='softmax'))
    return model


    

In [20]:

    

test = pd.read_csv('../data/GTSRB/GT-final_test.csv',sep=';')

X_test = []
y_test = []
i = 0
for file_name, class_id  in zip(list(test['Filename']), list(test['ClassId'])):
    img_path = os.path.join('/Users/mohammedamarnah/Desktop/SDCProject/CBM64-SDC/Traffic Recognition/data/GTSRB/Final_Test/Images/',file_name)
    X_test.append(preprocess_img(io.imread(img_path)))
    y_test.append(class_id)
    
X_test = np.array(X_test)
y_test = np.array(y_test)


    

    




//anaconda/envs/carnd-term1/lib/python3.5/site-packages/skimage/transform/_warps.py:84: UserWarning: The default mode, 'constant', will be changed to 'reflect' in skimage 0.15.
  warn("The default mode, 'constant', will be changed to 'reflect' in "

In [21]:

    

model = cnn_model()
model.load_weights('model.h5')
lr = 0.01
sgd = SGD(lr=lr, decay=1e-6, momentum=0.9, nesterov=True)
model.compile(loss='sparse_categorical_crossentropy',
          optimizer=sgd,
          metrics=['accuracy'])


    

In [47]:

    

print(X_test[0].shape)
img = np.reshape(X_test[0], (X_test[0].shape[1], X_test[0].shape[2], X_test[0].shape[0]))
print(img.shape)
img = np.asarray(img)
plt.imshow(img)


    

    




(3, 48, 48)
(48, 48, 3)






    
Out[47]:





<matplotlib.image.AxesImage at 0x14dc3c6a0>

In [50]:

    

score = model.predict(np.expand_dims(X_test[0],axis=0), batch_size=1)
score = np.asarray(score)
mx = -1e9
idx = 0
for i in range(len(score[0])):
    if score[0][i] >= mx:
        mx = score[0][i]
        idx = i
print(i)
print(mx)


    

    




43
1.0

In [ ]: