def _get_batches_of_transformed_samples(self, index_array):
batch_x = np.zeros(shape=(len(index_array),) + self.target_size + (3,))
batch_y = np.zeros(shape=(len(index_array),) + self.target_size + (self.num_classes,))
for i, idx in enumerate(index_array):
image, label = load_image(self.images_list[idx]), load_image(self.labels_list[idx])
if self.image_data_generator.random_crop:
image, label = random_crop(image, label, self.target_size)
else:
image, label = resize_image(image, label, self.target_size)
if np.random.uniform(0., 1.) < self.data_aug_rate:
if np.random.randint(2):
image, label = random_vertical_flip(image, label, self.image_data_generator.vertical_flip)
if np.random.randint(2):
image, label = random_horizontal_flip(image, label, self.image_data_generator.horizontal_flip)
if np.random.randint(2):
image, label = random_brightness(image, label, self.image_data_generator.brightness_range)
if np.random.randint(2):
image, label = random_rotation(image, label, self.image_data_generator.rotation_range)
if np.random.randint(2):
image, label = random_channel_shift(image, label, self.image_data_generator.channel_shift_range)
if np.random.randint(2):
image, label = random_zoom(image, label, self.image_data_generator.zoom_range)
image = imagenet_utils.preprocess_input(image.astype('float32'), data_format='channels_last',
mode='torch')
label = one_hot(label, self.num_classes)
batch_x[i], batch_y[i] = image, label
return batch_x, batch_y
def create_features(dataset, pre_model):
x_scratch = []
i = 0
# loop over the images
print("loop started")
for imagePath in dataset:
print("image", i, "\n")
# load the input image and image is resized to 224x224 pixels
# 299x299 for inceptionV3
image = load_img(imagePath, target_size=(224, 224))
image = img_to_array(image)
# preprocess the image by (1) expanding the dimensions and
# (2) subtracting the mean RGB pixel intensity from the
# ImageNet dataset
image = np.expand_dims(image, axis=0)
image = imagenet_utils.preprocess_input(image)
# add the image to the batch
x_scratch.append(image)
i += 1
print("for loop done")
x = np.vstack(x_scratch)
features = pre_model.predict(x, batch_size=32)
# features_flatten = features.reshape((features.shape[0], 7 * 7 * 512))
return x, features
def _get_X_and_names(self, model: CnnModel, list_fams, num_samples,
property_alias: str) -> Tuple[np.ndarray, List[str]]:
"""
Used to get all the features from a given set along with their names
:param list_fams: the list of families
:param num_samples: the number of samples
:param property_alias: the property to be looked
:return: a Tuple containing a Numpy array with the features and a List containing the name of the images
"""
channels, width, height = model.input_shape
X_train = np.zeros((num_samples, width, height, channels))
cnt = 0
samples_names = []
print("Processing images ...")
for i in range(len(list_fams)):
print('current fam: ', i)
for index, img_file in enumerate(
self._fetch_all_images(join(list_fams[i],
property_alias))):
img = image.load_img(img_file, target_size=(width, height))
x = image.img_to_array(img)
x = np.expand_dims(x, axis=0)
x = preprocess_input(x)
X_train[cnt] = x
cnt += 1
index = img_file.find(self.frame_delimiter)
samples_names.append(img_file[0:index])
return X_train, samples_names
def preprocess(x, pre_process_method):
if pre_process_method not in dic_mine_to_keras.keys():
raise ValueError("mode {} doesn't supported. Expected values: {}".format(pre_process_method, dic_mine_to_keras.keys()))
if isinstance(x, np.ndarray):
t = deepcopy(x)
else:
t = x
return preprocess_input(x=t, mode=dic_mine_to_keras[pre_process_method], data_format='channels_last')
def preprocess_input(x, **kwargs):
"""Preprocesses a numpy array encoding a batch of images.
# Arguments
x: a 4D numpy array consists of RGB values within [0, 255].
# Returns
Preprocessed array.
"""
return imagenet_utils.preprocess_input(x, mode='tf', **kwargs)
def preprocess_input(x):
"""Preprocesses a numpy array encoding a batch of images.
# Arguments
x: a 4D numpy array consists of RGB values within [0, 255].
# Returns
Input array scaled to [-1.,1.]
"""
return imagenet_utils.preprocess_input(x, mode='tf')
def preprocess_input_keras(x, **kwargs):
"""Preprocesses a numpy array encoding a batch of images.
# Arguments
x: a 4D numpy array consists of RGB values within [0, 255].
data_format: data format of the image tensor.
# Returns
Preprocessed array.
"""
return preprocess_input(x, mode='torch', **kwargs)
def test_preprocess_input_symbolic():
# Test image batch
x = np.random.uniform(0, 255, (2, 10, 10, 3))
inputs = layers.Input(shape=x.shape[1:])
outputs = layers.Lambda(utils.preprocess_input,
output_shape=x.shape[1:])(inputs)
model = models.Model(inputs, outputs)
assert model.predict(x).shape == x.shape
outputs1 = layers.Lambda(
lambda x: utils.preprocess_input(x, 'channels_last'),
output_shape=x.shape[1:])(inputs)
model1 = models.Model(inputs, outputs1)
out1 = model1.predict(x)
x2 = np.transpose(x, (0, 3, 1, 2))
inputs2 = layers.Input(shape=x2.shape[1:])
outputs2 = layers.Lambda(
lambda x: utils.preprocess_input(x, 'channels_first'),
output_shape=x2.shape[1:])(inputs2)
model2 = models.Model(inputs2, outputs2)
out2 = model2.predict(x2)
assert_allclose(out1, out2.transpose(0, 2, 3, 1))
# Test single image
x = np.random.uniform(0, 255, (10, 10, 3))
inputs = layers.Input(shape=x.shape)
outputs = layers.Lambda(utils.preprocess_input,
output_shape=x.shape)(inputs)
model = models.Model(inputs, outputs)
assert model.predict(x[np.newaxis])[0].shape == x.shape
outputs1 = layers.Lambda(
lambda x: utils.preprocess_input(x, 'channels_last'),
output_shape=x.shape)(inputs)
model1 = models.Model(inputs, outputs1)
out1 = model1.predict(x[np.newaxis])[0]
x2 = np.transpose(x, (2, 0, 1))
inputs2 = layers.Input(shape=x2.shape)
outputs2 = layers.Lambda(
lambda x: utils.preprocess_input(x, 'channels_first'),
output_shape=x2.shape)(inputs2)
model2 = models.Model(inputs2, outputs2)
out2 = model2.predict(x2[np.newaxis])[0]
assert_allclose(out1, out2.transpose(1, 2, 0))
def preprocess_input(rgb_values, data_format=None):
"""Preprocesses a numpy array encoding a batch of images.
# Arguments
rgb_values: a 3D or 4D numpy array consists of RGB values within [0, 255].
data_format: data format of the image tensor.
# Returns
Preprocessed array.
"""
return imagenet_utils.preprocess_input(rgb_values, data_format, mode='torch')
def read_multi_image(image_paths):
list_images = []
for (j, imagePath) in enumerate(image_paths):
print("reading image " + str(j))
img = load_img(imagePath, target_size=(224, 224))
img = img_to_array(img)
img = np.expand_dims(img, 0)
img = imagenet_utils.preprocess_input(img)
list_images.append(img)
list_images = np.vstack(list_images)
return list_images
def prepare_image(image, target):
# image mode should be "RGB"
if image.mode != "RGB":
image = image.convert("RGB")
# resize for model
image = image.resize(target)
image = img_to_array(image)
image = np.expand_dims(image, axis=0)
image = imagenet_utils.preprocess_input(image)
# return it
return image
def _get_batches_of_transformed_samples(self, index_array):
batch_x = []
batch_y = []
for batch_index, image_index in enumerate(index_array):
id = self.image_ids[image_index]
img_name = self.image_name_template.format(id=id)
path = os.path.join(self.images_dir, img_name)
image = np.array(img_to_array(load_img(path)), "uint8")
mask_name = self.mask_template.format(id=id)
mask_path = os.path.join(self.masks_dir, mask_name)
mask = cv2.imread(mask_path, cv2.IMREAD_COLOR)
label = cv2.imread(
os.path.join(self.labels_dir,
self.label_template.format(id=id)),
cv2.IMREAD_UNCHANGED)
if args.use_full_masks:
mask[..., 0] = (label > 0) * 255
if self.crop_shape is not None:
crop_mask, crop_image, crop_label = self.augment_and_crop_mask_image(
mask, image, label, id, self.crop_shape)
data = self.random_transformer(
image=np.array(crop_image, "uint8"),
mask=np.array(crop_mask, "uint8"))
crop_image, crop_mask = data['image'], data['mask']
if len(np.shape(crop_mask)) == 2:
crop_mask = np.expand_dims(crop_mask, -1)
crop_mask = self.transform_mask(crop_mask, crop_image)
batch_x.append(crop_image)
batch_y.append(crop_mask)
else:
x0, x1, y0, y1 = 0, 0, 0, 0
if (image.shape[1] % 32) != 0:
x0 = int((32 - image.shape[1] % 32) / 2)
x1 = (32 - image.shape[1] % 32) - x0
if (image.shape[0] % 32) != 0:
y0 = int((32 - image.shape[0] % 32) / 2)
y1 = (32 - image.shape[0] % 32) - y0
image = np.pad(image, ((y0, y1), (x0, x1), (0, 0)), 'reflect')
mask = np.pad(mask, ((y0, y1), (x0, x1), (0, 0)), 'reflect')
batch_x.append(image)
mask = self.transform_mask(mask, image)
batch_y.append(mask)
batch_x = np.array(batch_x, dtype="float32")
batch_y = np.array(batch_y, dtype="float32")
if self.preprocessing_function:
batch_x = imagenet_utils.preprocess_input(
batch_x, mode=self.preprocessing_function)
return self.transform_batch_x(batch_x), self.transform_batch_y(batch_y)
def __init__(self, config):
"""Constructor.
"""
self.config = config
idxs = []
with open(self.config['path'] + '/seq_ref_clean.txt',
'r',
encoding="latin-1") as f:
for l, line in enumerate(f):
if l >= 4:
idxs.append([int(j) for j in line.split(' ')])
idxs[-1][1] = 4371
X = self.config['pad_value'] * np.ones(
(len(idxs), np.max(np.diff(np.array(idxs), axis=1)) + 1,
self.config['size'], self.config['size'], 3),
dtype=self.config['dtype'])
y = np.zeros((len(idxs), ), dtype='int32')
for i, idx in enumerate(idxs):
for j, q in enumerate(range(idx[0], idx[1] + 1)):
img = imread(self.config['path'] + '/img_clean/img' +
'{:04d}'.format(q) + '.jpg')
img = imresize(img, (self.config['size'], self.config['size']))
X[i, j, :, :, :] = img
y[i] = i
X /= 255.
# zca
if self.config['zca_epsilon'] is not None:
self.ZCA = ZCAWhitening(epsilon=self.config['zca_epsilon'])
X = self.ZCA.fit_transform(X)
if self.config['imagenet_preprocessing'] is True:
X = np.pad(X, ((0, 0), (0, 0), (0, 32 - self.config['size']),
(0, 32 - self.config['size']), (0, 0)), 'constant')
for i in range(len(X)):
for t in range(len(X[0])):
X[i, t] = imresize(X[i, t], (32, 32))
X = preprocess_input(X, data_format='channels_last')
Y = to_categorical(y, len(X))
self.X = X
self.Y = Y
print(self.X.shape, self.Y.shape)
def __getitem__(self, idx):
batch_start = idx * self.config['batch_size']
batch_end = np.min(
[batch_start + self.config['batch_size'],
len(self.videos)])
metas = self.metas[batch_start:batch_end]
lens = [v[0] for v in metas]
X = []
for v, video in enumerate(self.videos[batch_start:batch_end]):
frame_start = np.random.randint(lens[v] - self.config['frames'])
frame_end = np.min([frame_start + self.config['frames'], lens[v]])
frames = self.build_frames(video,
metas[v],
frame_start=frame_start,
frame_end=frame_end)
X.append(frames)
X = np.array(X).astype(self.config['dtype'])
if self.config['zca_epsilon'] is not None:
if self.precomputing is True:
X = self.zca.partial_fit_transform(X)
else:
if self.train_val_test in [
'train', 'train_labeled'
] and 'size_crop' in self.config.keys():
X = self.zca.partial_fit_transform(X)
else:
X = self.zca.transform(X)
if self.config['imagenet_preprocessing'] is True:
if X.shape[-1] == 1:
X = np.repeat(X, 3, axis=-1)
X = np.pad(X, ((0, 0), (0, 0), (0, 32 - self.config['size']),
(0, 32 - self.config['size']), (0, 0)), 'constant')
for i in range(len(X)):
for t in range(len(X[0])):
X[i, t] = imresize(X[i, t], (32, 32))
X = preprocess_input(X, data_format='channels_last')
if self.classes is None:
Y = X
else:
Y = []
for v, video in enumerate(self.videos[batch_start:batch_end]):
y = self.classes.index(video.split('/')[0])
Y.append(y)
Y = to_categorical(Y, len(self.classes))
return (X, Y)
def __init__(self, config):
"""Constructor.
"""
self.config = config
X = np.zeros((100, 72, self.config['size'], self.config['size'], 3),
dtype=self.config['dtype'])
y = np.zeros((100, ), dtype='int32')
for i in range(len(X)):
for j, t in enumerate(range(0, 360, 5)):
img = imread(self.config['path'] + '/obj' + str(i + 1) + '__' +
str(t) + '.png')
img = imresize(img, (self.config['size'], self.config['size']))
X[i, j, :, :, :] = img
y[i] = i
X /= 255.
# zca
if self.config['zca_epsilon'] is not None:
self.ZCA = ZCAWhitening(epsilon=self.config['zca_epsilon'])
X = self.ZCA.fit_transform(X)
if self.config['imagenet_preprocessing'] is True:
X = np.pad(X, ((0, 0), (0, 0), (0, 32 - self.config['size']),
(0, 32 - self.config['size']), (0, 0)), 'constant')
for i in range(len(X)):
for t in range(len(X[0])):
X[i, t] = imresize(X[i, t], (32, 32))
X = preprocess_input(X, data_format='channels_last')
Y = to_categorical(y, len(X))
self.X = X
self.Y = Y
print(self.X.shape, self.Y.shape)
maxpool = model.get_layer(name='block5_pool')
shape = maxpool.output_shape[1:]
dense = model.get_layer(name='fc1')
layer_utils.convert_dense_weights_data_format(
dense, shape, 'channels_first')
if K.backend() == 'tensorflow':
warnings.warn('You are using the TensorFlow backend, yet you '
'are using the Theano '
'image data format convention '
'(`image_data_format="channels_first"`). '
'For best performance, set '
'`image_data_format="channels_last"` in '
'your Keras config '
'at ~/.keras/keras.json.')
return model
if __name__ == '__main__':
model = VGG19(include_top=True, weights='imagenet')
img_path = 'cat.jpg'
img = image.load_img(img_path, target_size=(224, 224))
x = image.img_to_array(img)
x = np.expand_dims(x, axis=0)
x = preprocess_input(x)
print('Input image shape:', x.shape)
preds = model.predict(x)
print('Predicted:', decode_predictions(preds))
def preprocess_input(x, **kwargs):
return imagenet_utils.preprocess_input(x, mode='torch', **kwargs)
if args.color_csv is None:
csv_file = os.path.join('CamVid', 'class_dict.csv')
else:
csv_file = args.csv_file
_, color_values = get_colored_info(csv_file)
for i, name in enumerate(image_names):
sys.stdout.write('\rRunning test image %d / %d' %
(i + 1, len(image_names)))
sys.stdout.flush()
image = cv2.resize(load_image(name),
dsize=(args.crop_width, args.crop_height))
image = imagenet_utils.preprocess_input(image.astype(np.float32),
data_format='channels_last',
mode='torch')
# image processing
if np.ndim(image) == 3:
image = np.expand_dims(image, axis=0)
assert np.ndim(image) == 4
# get the prediction
prediction = net.predict(image)
if np.ndim(prediction) == 4:
prediction = np.squeeze(prediction, axis=0)
# decode one-hot
prediction = decode_one_hot(prediction)
def preprocess_input(*args, **kwargs):
return imagenet_utils.preprocess_input(*args, **kwargs)
def preprocess_input(*args, **kwargs):
kwargs['backend'] = backend
return utils.preprocess_input(*args, **kwargs)
# load data and transform gray images to rgb images
images = np.stack(
[gray2rgb(io.imread(image_dir + '/' + fn)) for fn in file_names], axis=0)
gt_maps = np.stack([io.imread(gt_dir + '/' + fn) for fn in file_names], axis=0)
# Zero padding the input images
pad_widths = [(0, 0), (8, 8), (12, 12),
(0, 0)] if dataset == 'cyc2_1488x1512' else [(0, 0), (0, 0),
(8, 8), (0, 0)]
images_pad = np.pad(images, pad_widths, mode='reflect')
## Image preprocessing
print('Preprocessing ...')
preprocess_input = cm.get_preprocessing(backbone) ## preprocessing function
images_pad = preprocess_input(images_pad)
# will scale pixels between 0 and 1 and then will normalize each channel with respect to the ImageNet dataset
print('Preprocessing done !')
## Load the trained model
model = tf.keras.models.load_model(model_folder + '/ready_model.h5')
## Label map prediction
pr_masks = model.predict(images_pad, batch_size=1)
## probability maps [N(num of images) x H x W x C(class)] for 0: G1, 1: S, 2: G2, 3: background/M phase
# crop images back
pr_masks = pr_masks[:, pad_widths[1][0]:-pad_widths[1][1], pad_widths[2][
0]:-pad_widths[2][
1], :] if dataset == 'cyc2_1488x1512' else pr_masks[:, :, pad_widths[
2][0]:-pad_widths[2][1], :]
pr_masks_ = np.zeros(pr_masks.shape, dtype=np.float32)
from keras.applications.resnet50 import ResNet50
from keras.preprocessing import image
from keras_applications.imagenet_utils import preprocess_input
from keras.applications.resnet50 import preprocess_input, decode_predictions
import numpy as np
model = ResNet50(weights='imagenet')
img_path = '/Users/mosampatel/Documents/mosam.jpeg'
img = image.load_img(img_path, target_size=(224, 224))
x = image.img_to_array(img)
x = np.expand_dims(x, axis=0)
x = preprocess_input(x)
preds = model.predict(x)
# decode the results into a list of tuples (class, description, probability)
# (one such list for each sample in the batch)
print('Predicted:', decode_predictions(preds, top=3)[0])
def __init__(self, config):
"""Constructor.
"""
self.config = config
(X_train, y_train), (X_test, y_test) = cifar10.load_data()
X_train = X_train.reshape(-1, 32, 32, 3)
X_test = X_test.reshape(-1, 32, 32, 3)
X_train = X_train.astype(self.config['dtype'])
X_test = X_test.astype(self.config['dtype'])
N = ((32 - self.config['size']) // self.config['stride']) + 1
X_train_scan = np.zeros((X_train.shape[0], N**2, self.config['size'],
self.config['size'], 3))
X_test_scan = np.zeros((X_test.shape[0], N**2, self.config['size'],
self.config['size'], 3))
for i in range(len(X_train)):
for t1 in range(N):
for t2 in (range(N) if
(np.mod(t1, 2) == 0) else range(N - 1, -1, -1)):
X_train_scan[i, t1 * N +
(t2 if (np.mod(t1, 2) == 0) else
(N - 1 - t2)), :, :, :] = X_train[i, (
t2 * self.config['stride']):(
t2 * self.config['stride'] +
self.config['size']), (
t1 * self.config['stride']):(
t1 * self.config['stride'] +
self.config['size']), :]
for i in range(len(X_test)):
for t1 in range(N):
for t2 in (range(N) if
(np.mod(t1, 2) == 0) else range(N - 1, -1, -1)):
X_test_scan[i,
t1 * N + (t2 if (np.mod(t1, 2) == 0) else
(N - 1 - t2)), :, :, :] = X_test[
i, (t2 * self.config['stride']):(
t2 * self.config['stride'] +
self.config['size']),
(t1 * self.config['stride']):(
t1 * self.config['stride'] +
self.config['size']), :]
X_train_scan /= 255.
X_test_scan /= 255.
# zca
if self.config['zca_epsilon'] is not None:
self.ZCA = ZCAWhitening(epsilon=self.config['zca_epsilon'])
X_train_scan = self.ZCA.fit_transform(X_train_scan)
X_test_scan = self.ZCA.transform(X_test_scan)
if self.config['imagenet_preprocessing'] is True:
X_train_scan = np.pad(X_train_scan,
((0, 0), (0, 0),
(0, 32 - self.config['size']),
(0, 32 - self.config['size']), (0, 0)),
'constant')
X_test_scan = np.pad(X_test_scan,
((0, 0), (0, 0),
(0, 32 - self.config['size']),
(0, 32 - self.config['size']), (0, 0)),
'constant')
for i in range(len(X_train_scan)):
for t in range(len(X_train_scan[0])):
X_train_scan[i, t] = imresize(X_train_scan[i, t], (32, 32))
for i in range(len(X_test_scan)):
for t in range(len(X_test_scan[0])):
X_test_scan[i, t] = imresize(X_test_scan[i, t], (32, 32))
X_train_scan = preprocess_input(X_train_scan,
data_format='channels_last')
X_test_scan = preprocess_input(X_test_scan,
data_format='channels_last')
nb_classes = 10
Y_train = to_categorical(y_train, nb_classes)
Y_test = to_categorical(y_test, nb_classes)
self.X_train = X_train_scan
self.Y_train = Y_train
self.X_test = X_test_scan
self.Y_test = Y_test
print(self.X_train.shape, self.Y_train.shape, self.X_test.shape,
self.Y_test.shape)
def __init__(self, config):
"""Constructor.
"""
self.config = config
(X_train, y_train), (X_test, y_test) = mnist.load_data()
X_train = X_train.reshape(-1, 28, 28, 1)
X_test = X_test.reshape(-1, 28, 28, 1)
X_train = X_train.astype(self.config['dtype'])
X_test = X_test.astype(self.config['dtype'])
X_train_rot = np.zeros((X_train.shape[0], self.config['rotations'], self.config['size'], self.config['size'], 1))
X_test_rot = np.zeros((X_test.shape[0], self.config['rotations'], self.config['size'], self.config['size'], 1))
for i in range(len(X_train)):
img = Image.fromarray(np.reshape(X_train[i, :, :, :], (28, 28)))
img = img.convert('L')
phase = 360. * np.random.rand()
for t in range(self.config['rotations']):
_img = img.rotate(phase + t * 360. / self.config['rotations'], Image.BILINEAR).resize((28, 28), Image.BILINEAR).getdata()
_img = np.reshape(np.array(_img, dtype=np.uint8), (28, 28, 1))[:self.config['size'], :self.config['size'], :]
X_train_rot[i, t, :, :, :] = _img.astype(dtype=self.config['dtype'])
for i in range(len(X_test)):
img = Image.fromarray(np.reshape(X_test[i, :, :, :], (28, 28)))
img = img.convert('L')
phase = 360. * np.random.rand()
for t in range(self.config['rotations']):
_img = img.rotate(phase + t * 360. / self.config['rotations'], Image.BILINEAR).resize((28, 28), Image.BILINEAR).getdata()
_img = np.reshape(np.array(_img, dtype=np.uint8), (28, 28, 1))[:self.config['size'], :self.config['size'], :]
X_test_rot[i, t, :, :, :] = _img.astype(dtype=self.config['dtype'])
X_train_rot /= 255.
X_test_rot /= 255.
# zca
if self.config['zca_epsilon'] is not None:
self.ZCA = ZCAWhitening(epsilon=self.config['zca_epsilon'])
X_train_rot = self.ZCA.fit_transform(X_train_rot)
X_test_rot = self.ZCA.transform(X_test_rot)
if self.config['imagenet_preprocessing'] is True:
X_train_rot = np.pad(X_train_rot, ((0, 0), (0, 0), (0, 32 - self.config['size']), (0, 32 - self.config['size']), (0, 0)), 'constant')
X_test_rot = np.pad(X_test_rot, ((0, 0), (0, 0), (0, 32 - self.config['size']), (0, 32 - self.config['size']), (0, 0)), 'constant')
X_train_rot = np.repeat(X_train_rot, 3, axis=-1)
X_test_rot = np.repeat(X_test_rot, 3, axis=-1)
for i in range(len(X_train_rot)):
for t in range(len(X_train_rot[0])):
X_train_rot[i, t] = imresize(X_train_rot[i, t], (32, 32))
for i in range(len(X_test_rot)):
for t in range(len(X_test_rot[0])):
X_test_rot[i, t] = imresize(X_test_rot[i, t], (32, 32))
X_train_rot = preprocess_input(X_train_rot, data_format='channels_last')
X_test_rot = preprocess_input(X_test_rot, data_format='channels_last')
nb_classes = 10
Y_train = to_categorical(y_train, nb_classes)
Y_test = to_categorical(y_test, nb_classes)
self.X_train = X_train_rot
self.Y_train = Y_train
self.X_test = X_test_rot
self.Y_test = Y_test
print(self.X_train.shape, self.Y_train.shape, self.X_test.shape, self.Y_test.shape)
def preprocessing_function_caffe(x):
return preprocess_input(x, data_format='channels_last', mode='caffe')
def preprocess_inputs(x):
return preprocess_input(x, mode=args.preprocessing_function)
def preprocess_input(*args, **kwargs):
return imagenet_utils.preprocess_input(*args, **kwargs)
def test_preprocess_input():
# Test image batch with float and int image input
x = np.random.uniform(0, 255, (2, 10, 10, 3))
xint = x.astype('int32')
assert utils.preprocess_input(x).shape == x.shape
assert utils.preprocess_input(xint).shape == xint.shape
out1 = utils.preprocess_input(x, 'channels_last')
out1int = utils.preprocess_input(xint, 'channels_last')
out2 = utils.preprocess_input(np.transpose(x, (0, 3, 1, 2)),
'channels_first')
out2int = utils.preprocess_input(np.transpose(xint, (0, 3, 1, 2)),
'channels_first')
assert_allclose(out1, out2.transpose(0, 2, 3, 1))
assert_allclose(out1int, out2int.transpose(0, 2, 3, 1))
# Test single image
x = np.random.uniform(0, 255, (10, 10, 3))
xint = x.astype('int32')
assert utils.preprocess_input(x).shape == x.shape
assert utils.preprocess_input(xint).shape == xint.shape
out1 = utils.preprocess_input(x, 'channels_last')
out1int = utils.preprocess_input(xint, 'channels_last')
out2 = utils.preprocess_input(np.transpose(x, (2, 0, 1)), 'channels_first')
out2int = utils.preprocess_input(np.transpose(xint, (2, 0, 1)),
'channels_first')
assert_allclose(out1, out2.transpose(1, 2, 0))
assert_allclose(out1int, out2int.transpose(1, 2, 0))
# Test that writing over the input data works predictably
for mode in ['torch', 'tf']:
x = np.random.uniform(0, 255, (2, 10, 10, 3))
xint = x.astype('int')
x2 = utils.preprocess_input(x, mode=mode)
xint2 = utils.preprocess_input(xint)
assert_allclose(x, x2)
assert xint.astype('float').max() != xint2.max()
# Caffe mode works differently from the others
x = np.random.uniform(0, 255, (2, 10, 10, 3))
xint = x.astype('int')
x2 = utils.preprocess_input(x, data_format='channels_last', mode='caffe')
xint2 = utils.preprocess_input(xint)
assert_allclose(x, x2[..., ::-1])
assert xint.astype('float').max() != xint2.max()
