def main():
tarpath = "cifar-10-python.tar.gz"
data_path = extract(tarpath)
images, labels = load_batch(data_path[0])
labels_strings = load_meta()
fig = plt.figure()
N = 16
for i, img, label_i in zip(range(N), images[:N], labels[:N]):
img = np.reshape(img, [32, 32, 3], order="F")
img = np.rot90(img, k=3)
ax = fig.add_subplot(4, 4, i + 1)
ax.imshow(img)
ax.set_xticklabels([])
ax.set_yticklabels([])
ax.axis('off')
ax.set_title(labels_strings[label_i])
plt.show()
def compute_hog_features(images):
features = []
hog_images = []
for image in images:
fd, hog_image = hog(image, visualize=True)
features.append(fd)
hog_images.append(hog_image)
return np.array(features), hog_images
if __name__ == '__main__':
# Import data
X_train, y_train = load_batch()
X_test, y_test = load_batch(test=True)
# Subset data to 1/10th
n_train = X_train.shape[0] // 2
X_train = X_train[0:n_train]
y_train = y_train[0:n_train]
print("Number of training images:", n_train)
# Number of training images: 5000
n_test = X_test.shape[0] // 10
X_test = X_test[0:n_test]
y_test = y_test[0:n_test]
print("Number of testing images:", n_test)
# Number of testing images: 1000
from load_cifar import load_batch
import numpy as np
import matplotlib.pyplot as plt
X_train, y_train = load_batch()
n = 0
fig = plt.figure()
for c in range(0, 10):
# Find images in class c
idxs = np.where(y_train == c)[0][0:10]
# Add them to the plot
for i in idxs:
a = fig.add_subplot(10, 10, n + 1)
plt.imshow(X_train[i, :, :, :])
plt.axis("off")
n += 1
fig.set_size_inches(12, 12)
plt.savefig('cifar10test.png')
from load_cifar import load_batch
import numpy as np
images, _ = load_batch()
images = np.reshape(images, (10000, 3, 32, 32))
from scipy.misc import imsave
imsave('cat.jpg', images[26]) # image no #