def load_images(path= "/fashion_mnist/data/fashion", kind='train'):
mnist_path = os.getcwd() + path
labels = ['t_shirt_top', 'trouser', 'pullover', 'dress', 'coat', 'sandal', 'shirt', 'sneaker', 'bag', 'ankle_boots']
# reading the data using the utility function load_mnist
# train_images, train_labels = load_mnist(mnist_path, kind='train')
# t10k_images, t10k_labels = load_mnist(mnist_path, kind='t10k')
# train_labels_str = image_class_to_str(train_labels)
# t10k_labels_str = image_class_to_str(t10k_labels)
return load_mnist(mnist_path, kind=kind)
from fashion_mnist.utils.mnist_reader import load_mnist
import numpy as np
import statistics as stat
import matplotlib.pyplot as plt
import math
import pickle
image, label = load_mnist('./fashion_mnist/data/mnist/')
all_img = image.copy()
all_lbl = label.copy()
t_images, t_labels = load_mnist('./fashion_mnist/data/mnist/', kind='t10k')
all_test_images = t_images.copy()
all_test_labels = t_labels.copy()
digit_type = [1, 8] # pos 1, negt 8
N = 2 # no. of classes
cm_2 = np.zeros((N, N))
classes = list(range(N))
unq, cnt = np.unique(
all_lbl,
return_counts=True) # no. of data points of a class in training set
unq1, cnt1 = np.unique(
all_test_labels,
return_counts=True) # no. of data points in test set of the same class
print(unq, cnt)
print(unq1, cnt1)
class_priors = cnt / len(all_lbl) # cnt : No. of data points of each digit
total_train_points = cnt[digit_type[0]] + cnt[digit_type[1]]
def load_mnist_dataset():
dir_loc = '/home/kamil/PycharmProjects/image-recognition-msid/fashion_mnist/data/fashion'
train_set = mnist_reader.load_mnist(dir_loc, kind='train')
test_set = mnist_reader.load_mnist(dir_loc, kind='t10k')
return train_set, test_set
def get_data():
"""
Get data ready to learn with.
Returns
-------
dict
"""
simple = False
if simple:
from sklearn.datasets import load_digits
from sklearn.utils import shuffle
digits = load_digits()
x = [np.array(el).flatten() for el in digits.images]
y = digits.target
x = np.divide(x, 255.0) * 2 - 1
x, y = shuffle(x, y, random_state=0)
from sklearn.cross_validation import train_test_split
x_train, x_test, y_train, y_test = train_test_split(x,
y,
test_size=0.33,
random_state=42)
data = {
'train': {
'X': x_train,
'y': y_train
},
'test': {
'X': x_test,
'y': y_test
}
}
else:
# from sklearn.datasets import fetch_mldata
from sklearn.utils import shuffle
# mnist = fetch_mldata("MNIST original")
# x = mnist.data
# y = mnist.target
# x = x / 255.0 * 2 - 1
#
# x, y = shuffle(x, y, random_state=0)
#
# from sklearn.cross_validation import train_test_split
#
# x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=0.33, random_state=42)
X_train, y_train = m_reader.load_mnist('fashion_mnist/data/fashion',
kind='train')
X_test, y_test = m_reader.load_mnist('fashion_mnist/data/fashion',
kind='t10k')
X_test = X_test / 255.0 * 2 - 1
X_train = X_train / 255.0 * 2 - 1
data = {
'train': {
'X': X_train,
'y': y_train
},
'test': {
'X': X_test,
'y': y_test
}
}
return data
from vis.visualization import visualize_activation
from vis.utils import utils
config=tf.ConfigProto()
config.gpu_options.allow_growth = True
set_session(tf.Session(config=config))
batch_size = 400
num_classes = 10
epochs = 20
# input image dimensions
img_rows, img_cols = 28, 28
# the data, shuffled and split between train and test sets
x_train, y_train = mnist_reader.load_mnist('fashion_mnist/data/fashion', kind='train')
x_test, y_test = mnist_reader.load_mnist('fashion_mnist/data/fashion', kind='t10k')
if K.image_data_format() == 'channels_first':
x_train = x_train.reshape(x_train.shape[0], 1, img_rows, img_cols)
x_test = x_test.reshape(x_test.shape[0], 1, img_rows, img_cols)
input_shape = (1, img_rows, img_cols)
else:
x_train = x_train.reshape(x_train.shape[0], img_rows, img_cols, 1)
x_test = x_test.reshape(x_test.shape[0], img_rows, img_cols, 1)
input_shape = (img_rows, img_cols, 1)
x_train = x_train.astype('float32')
x_test = x_test.astype('float32')
x_train /= 255
x_test /= 255
from io import BytesIO
from timeit import default_timer as timer
import numpy as np
import tmap as tm
from faerun import Faerun
from PIL import Image
from fashion_mnist.utils import mnist_reader
# Coniguration for the tmap layout
CFG = tm.LayoutConfiguration()
CFG.node_size = 1 / 55
# Load fashion mnist data
IMAGES_TRAIN, LABELS_TRAIN = mnist_reader.load_mnist(
"fashion_mnist/data/fashion", kind="train")
IMAGES_TEST, LABELS_TEST = mnist_reader.load_mnist(
"fashion_mnist/data/fashion", kind="t10k")
IMAGES = np.concatenate((IMAGES_TRAIN, IMAGES_TEST))
LABELS = np.concatenate((LABELS_TRAIN, LABELS_TEST))
IMAGE_LABELS = []
def main():
""" Main function """
# Initialize and configure tmap
dims = 1024
enc = tm.Minhash(28 * 28, 42, dims)
lf = tm.LSHForest(dims * 2, 128)
learning_rate = 7e-3
batch_size = 256
n_epochs = 100
n_train = None
n_test = None
lamda = 0
# Step 1: Read in data
fmnist_folder = './fashion_mnist'
#Create dataset load function [Refer fashion mnist github page for util function]
#Create train,validation,test split
#train, val, test = utils.read_fmnist(fmnist_folder, flatten=True)
# Step 2: Create datasets and iterator
# create training Dataset and batch it
x_train,y_train = load_mnist(os.path.join("fashion_mnist","data","fashion"),"train")
x_train=x_train/255.0
x_train, x_val, y_train, y_val = train_test_split(x_train, y_train, test_size=0.2, random_state=random_seed)
print(x_train.shape,y_train.shape)
n_train = x_train.shape[0]
print(x_val.shape,y_val.shape)
# create testing Dataset and batch it
x_test,y_test = load_mnist(os.path.join("fashion_mnist","data","fashion"),"t10k")
x_test=x_test/255.0
print(x_test.shape,y_test.shape)
n_test = x_test.shape[0]
